Flu season is coming—and so is the risk of an all-new bird flu

This article first appeared in The Checkup, MIT Technology Review’s weekly biotech newsletter. To receive it in your inbox every Thursday, and read articles like this first, sign up here.

September will soon be drawing to a close. The kids are back to school, and those of us in the Northern Hemisphere are experiencing the joys the end of summer brings: the cooling temperatures, the falling leaves, and, inevitably, the start of flu season.

I was reminded of that fact when my littlest woke me for an early-morning cuddle, sneezed into my face, and wiped her nose on my pajamas. I booked her flu vaccine the next morning.

In the US, the Centers for Disease Control and Prevention recommends the flu vaccine for everyone over six months old. This year, following the spread of the “bird flu” H5N1 in cattle, the CDC is especially urging dairy farm workers to get vaccinated. At the end of July, the organization announced a $10 million plan to deliver free flu shots to people who work with livestock.

The goal is not only to protect those workers from seasonal flu, but to protect us all from a potentially more devastating consequence: the emergence of a new form of flu that could trigger another pandemic. That hasn’t happened yet, but unfortunately, it’s looking increasingly possible.

First, it’s worth noting that flu viruses experience subtle changes in their genetic makeup all the time. This allows the virus to evolve rapidly, and it is why flu vaccines need to be updated every year, depending on which form of the virus is most likely to be circulating.

More dramatic genetic changes can take place when multiple flu viruses infect a single animal. The genome of a flu virus is made up of eight segments. When two different viruses end up in the same cell, they can swap segments with each other.

These swapping events can create all-new viruses. It’s impossible to predict exactly what will result, but there’s always a chance that the new virus will be easily spread or cause more serious disease than either of its predecessors.

The fear is that farm workers who get seasonal flu could also pick up bird flu from cows. Those people could become unwitting incubators for deadly new flu strains and end up passing them on to the people around them. “That is exactly how we think pandemics start,” says Thomas Peacock, a virologist at the Pirbright Institute in Woking, UK.

The virus responsible for the 2009 swine flu pandemic is thought to have come about this way. Its genome suggested it had resulted from the genetic reassortment of a mix of flu viruses, including some thought to largely infect pigs and others that originated in birds. Viruses with genes from both a human flu and a bird flu are thought to have been responsible for pandemics in 1918, 1957, and 1968, too.

The CDC is hoping that vaccinating these individuals against seasonal flu might lower the risk of history repeating. But unfortunately, it’s not an airtight solution. For a start, not everyone will get vaccinated. Around 45% of US agricultural workers are undocumented migrants, a group that tends to have low vaccination rates

Even if every farm worker were to be vaccinated, not all of them would be fully protected against getting sick with flu. The flu vaccine used in the US in 2019-2020 was 39% effective, but the one used in the 2004-2005 flu season was only 10% effective.

“It’s not a bad idea, but I don’t think it can get anywhere close to mitigating the underlying risk,” says Peacock.

I last reported on bird flu in February 2023. Back then, the virus was decimating bird populations, but there were no signs that it was making the jump to mammals, and it didn’t appear to be posing a risk to humans. “We don’t need to panic about a bird flu pandemic—yet,” was my conclusion at the time. Today, the picture is different. After speaking to virologists and scientists who are trying to track the spread of the current bird flu, I’ll admit that I am much more concerned about the potential for another pandemic.

The main advice for people who don’t work on farms is to avoid raw milk and dead animals, both of which could be harboring the virus. For the most part, we’re reliant on government agencies to monitor and limit the spread of this virus. And the limited actions that have been taken to date don’t exactly inspire much confidence.

“The barn door’s already open,” says Peacock. “This virus is already out and about.”


Now read the rest of The Checkup

Read more from MIT Technology Review’s archive

We don’t know how many dairy herds in the US are infected with H5N1 as the virus continues to spread. It could end up sticking around in farms forever, virologists told me earlier this week.

Manufacturing flu vaccines is a slow process that relies on eggs. But scientists hope mRNA flu vaccines could offer a quicker, cheaper, and more effective alternative.

Some flu vaccines are already made without eggs. One makes use of a virus synthesized in insect cells. Egg-free vaccines might even work better than those made using eggs, as Cassandra Willyard reported earlier this year.

Chickens are especially vulnerable to H5N1. Some scientists are exploring ways to edit the animals’ genes to make them more resilient to the virus, as Abdullahi Tsanni reported last year.

From around the web

Microplastics are everywhere. They even get inside our brains, possibly via our noses. (JAMA Network Open)

The majority of face transplants survive for at least 10 years, research has found. Of the 50 first face transplants, which were carried out across 11 countries, 85% survived for five years, and 74% for 10 years. (JAMA Surgery)

Don’t throw away that placenta! The organ holds clues to health and disease, and instead of being disposed of after birth, it should be carefully studied instead, scientists say. (Trends in Molecular Medicine)

In June, the drug lenacapavir was shown to be 100% effective at preventing HIV in women and adolescent girls. But while the drug was tested on women in Africa, it remains unavailable to most of them. (STAT)

We’re still getting to grips with what endometriosis is, how it works, and how to treat it. Women with the condition appear to have differences in their brain’s gray matter that can’t be explained by pelvic pain alone. (Human Reproduction)

AI models let robots carry out tasks in unfamiliar environments

It’s tricky to get robots to do things in environments they’ve never seen before. Typically, researchers need to train them on new data for every new place they encounter, which can become very time-consuming and expensive.

Now researchers have developed a series of AI models that teach robots to complete basic tasks in new surroundings without further training or fine-tuning. The five AI models, called robot utility models (RUMs), allow machines to complete five separate tasks—opening doors and drawers, and picking up tissues, bags, and cylindrical objects—in unfamiliar environments with a 90% success rate. 

The team, consisting of researchers from New York University, Meta, and the robotics company Hello Robot, hopes its findings will make it quicker and easier to teach robots new skills while helping them function within previously unseen domains. The approach could make it easier and cheaper to deploy robots in our homes.

“In the past, people have focused a lot on the problem of ‘How do we get robots to do everything?’ but not really asking ‘How do we get robots to do the things that they do know how to do—everywhere?’” says Mahi Shafiullah, a PhD student at New York University who worked on the project. “We looked at ‘How do you teach a robot to, say, open any door, anywhere?’”

Teaching robots new skills generally requires a lot of data, which is pretty hard to come by. Because robotic training data needs to be collected physically—a time-consuming and expensive undertaking—it’s much harder to build and scale training databases for robots than it is for types of AI like large language models, which are trained on information scraped from the internet.

To make it faster to gather the data essential for teaching a robot a new skill, the researchers developed a new version of a tool it had used in previous research: an iPhone attached to a cheap reacher-grabber stick, the kind typically used to pick up trash. 

The team used the setup to record around 1,000 demonstrations in 40 different environments, including homes in New York City and Jersey City, for each of the five tasks—some of which had been gathered as part of previous research. Then they trained learning algorithms on the five data sets to create the five RUM models.

These models were deployed on Stretch, a robot consisting of a wheeled unit, a tall pole, and a retractable arm holding an iPhone, to test how successfully they were able to execute the tasks in new environments without additional tweaking. Although they achieved a completion rate of 74.4%, the researchers were able to increase this to a 90% success rate when they took images from the iPhone and the robot’s head-mounted camera,  gave them to OpenAI’s recent GPT-4o LLM model, and asked it if the task had been completed successfully. If GPT-4o said no, they simply reset the robot and tried again.

A significant challenge facing roboticists is that training and testing their models in lab environments isn’t representative of what could happen in the real world, meaning research that helps machines to behave more reliably in new settings is much welcomed, says Mohit Shridhar, a research scientist specializing in robotic manipulation who wasn’t involved in the work. 

“It’s nice to see that it’s being evaluated in all these diverse homes and kitchens, because if you can get a robot to work in the wild in a random house, that’s the true goal of robotics,” he says.

The project could serve as a general recipe to build other utility robotics models for other tasks, helping to teach robots new skills with minimal extra work and making it easier for people who aren’t trained roboticists to deploy future robots in their homes, says Shafiullah.

“The dream that we’re going for is that I could train something, put it on the internet, and you should be able to download and run it on a robot in your home,” he says.

Why we need an AI safety hotline

In the past couple of years, regulators have been caught off guard again and again as tech companies compete to launch ever more advanced AI models. It’s only a matter of time before labs release another round of models that pose new regulatory challenges. We’re likely just weeks away, for example, from OpenAI’s release of ChatGPT-5, which promises to push AI capabilities further than ever before. As it stands, it seems there’s little anyone can do to delay or prevent the release of a model that poses excessive risks.

Testing AI models before they’re released is a common approach to mitigating certain risks, and it may help regulators weigh up the costs and benefits—and potentially block models from being released if they’re deemed too dangerous. But the accuracy and comprehensiveness of these tests leaves a lot to be desired. AI models may “sandbag” the evaluation—hiding some of their capabilities to avoid raising any safety concerns. The evaluations may also fail to reliably uncover the full set of risks posed by any one model. Evaluations likewise suffer from limited scope—current tests are unlikely to uncover all the risks that warrant further investigation. There’s also the question of who conducts the evaluations and how their biases may influence testing efforts. For those reasons, evaluations need to be used alongside other governance tools. 

One such tool could be internal reporting mechanisms within the labs. Ideally, employees should feel empowered to regularly and fully share their AI safety concerns with their colleagues, and they should feel those colleagues can then be counted on to act on the concerns. However, there’s growing evidence that, far from being promoted, open criticism is becoming rarer in AI labs. Just three months ago, 13 former and current workers from OpenAI and other labs penned an open letter expressing fear of retaliation if they attempt to disclose questionable corporate behaviors that fall short of breaking the law. 

How to sound the alarm

In theory, external whistleblower protections could play a valuable role in the detection of AI risks. These could protect employees fired for disclosing corporate actions, and they could help make up for inadequate internal reporting mechanisms. Nearly every state has a public policy exception to at-will employment termination—in other words, terminated employees can seek recourse against their employers if they were retaliated against for calling out unsafe or illegal corporate practices. However, in practice this exception offers employees few assurances. Judges tend to favor employers in whistleblower cases. The likelihood of AI labs’ surviving such suits seems particularly high given that society has yet to reach any sort of consensus as to what qualifies as unsafe AI development and deployment. 

These and other shortcomings explain why the aforementioned 13 AI workers, including ex-OpenAI employee William Saunders, called for a novel “right to warn.” Companies would have to offer employees an anonymous process for disclosing risk-related concerns to the lab’s board, a regulatory authority, and an independent third body made up of subject-matter experts. The ins and outs of this process have yet to be figured out, but it would presumably be a formal, bureaucratic mechanism. The board, regulator, and third party would all need to make a record of the disclosure. It’s likely that each body would then initiate some sort of investigation. Subsequent meetings and hearings also seem like a necessary part of the process. Yet if Saunders is to be taken at his word, what AI workers really want is something different. 

When Saunders went on the Big Technology Podcast to outline his ideal process for sharing safety concerns, his focus was not on formal avenues for reporting established risks. Instead, he indicated a desire for some intermediate, informal step. He wants a chance to receive neutral, expert feedback on whether a safety concern is substantial enough to go through a “high stakes” process such as a right-to-warn system. Current government regulators, as Saunders says, could not serve that role. 

For one thing, they likely lack the expertise to help an AI worker think through safety concerns. What’s more, few workers will pick up the phone if they know it’s a government official on the other end—that sort of call may be “very intimidating,” as Saunders himself said on the podcast. Instead, he envisages being able to call an expert to discuss his concerns. In an ideal scenario, he’d be told that the risk in question does not seem that severe or likely to materialize, freeing him up to return to whatever he was doing with more peace of mind. 

Lowering the stakes

What Saunders is asking for in this podcast isn’t a right to warn, then, as that suggests the employee is already convinced there’s unsafe or illegal activity afoot. What he’s really calling for is a gut check—an opportunity to verify whether a suspicion of unsafe or illegal behavior seems warranted. The stakes would be much lower, so the regulatory response could be lighter. The third party responsible for weighing up these gut checks could be a much more informal one. For example, AI PhD students, retired AI industry workers, and other individuals with AI expertise could volunteer for an AI safety hotline. They could be tasked with quickly and expertly discussing safety matters with employees via a confidential and anonymous phone conversation. Hotline volunteers would have familiarity with leading safety practices, as well as extensive knowledge of what options, such as right-to-warn mechanisms, may be available to the employee. 

As Saunders indicated, few employees will likely want to go from 0 to 100 with their safety concerns—straight from colleagues to the board or even a government body. They are much more likely to raise their issues if an intermediary, informal step is available.

Studying examples elsewhere

The details of how precisely an AI safety hotline would work deserve more debate among AI community members, regulators, and civil society. For the hotline to realize its full potential, for instance, it may need some way to escalate the most urgent, verified reports to the appropriate authorities. How to ensure the confidentiality of hotline conversations is another matter that needs thorough investigation. How to recruit and retain volunteers is another key question. Given leading experts’ broad concern about AI risk, some may be willing to participate simply out of a desire to lend a hand. Should too few folks step forward, other incentives may be necessary. The essential first step, though, is acknowledging this missing piece in the puzzle of AI safety regulation. The next step is looking for models to emulate in building out the first AI hotline. 

One place to start is with ombudspersons. Other industries have recognized the value of identifying these neutral, independent individuals as resources for evaluating the seriousness of employee concerns. Ombudspersons exist in academia, nonprofits, and the private sector. The distinguishing attribute of these individuals and their staffers is neutrality—they have no incentive to favor one side or the other, and thus they’re more likely to be trusted by all. A glance at the use of ombudspersons in the federal government shows that when they are available, issues may be raised and resolved sooner than they would be otherwise.

This concept is relatively new. The US Department of Commerce established the first federal ombudsman in 1971. The office was tasked with helping citizens resolve disputes with the agency and investigate agency actions. Other agencies, including the Social Security Administration and the Internal Revenue Service, soon followed suit. A retrospective review of these early efforts concluded that effective ombudspersons can meaningfully improve citizen-government relations. On the whole, ombudspersons were associated with an uptick in voluntary compliance with regulations and cooperation with the government. 

An AI ombudsperson or safety hotline would surely have different tasks and staff from an ombudsperson in a federal agency. Nevertheless, the general concept is worthy of study by those advocating safeguards in the AI industry. 

A right to warn may play a role in getting AI safety concerns aired, but we need to set up more intermediate, informal steps as well. An AI safety hotline is low-hanging regulatory fruit. A pilot made up of volunteers could be organized in relatively short order and provide an immediate outlet for those, like Saunders, who merely want a sounding board.

Kevin Frazier is an assistant professor at St. Thomas University College of Law and senior research fellow in the Constitutional Studies Program at the University of Texas at Austin.

Why OpenAI’s new model is such a big deal

This story is from The Algorithm, our weekly newsletter on AI. To get it in your inbox first, sign up here.

Last weekend, I got married at a summer camp, and during the day our guests competed in a series of games inspired by the show Survivor that my now-wife and I orchestrated. When we were planning the games in August, we wanted one station to be a memory challenge, where our friends and family would have to memorize part of a poem and then relay it to their teammates so they could re-create it with a set of wooden tiles. 

I thought OpenAI’s GPT-4o, its leading model at the time, would be perfectly suited to help. I asked it to create a short wedding-themed poem, with the constraint that each letter could only appear a certain number of times so we could make sure teams would be able to reproduce it with the provided set of tiles. GPT-4o failed miserably. The model repeatedly insisted that its poem worked within the constraints, even though it didn’t. It would correctly count the letters only after the fact, while continuing to deliver poems that didn’t fit the prompt. Without the time to meticulously craft the verses by hand, we ditched the poem idea and instead challenged guests to memorize a series of shapes made from colored tiles. (That ended up being a total hit with our friends and family, who also competed in dodgeball, egg tosses, and capture the flag.)    

However, last week OpenAI released a new model called o1 (previously referred to under the code name “Strawberry” and, before that, Q*) that blows GPT-4o out of the water for this type of purpose

Unlike previous models that are well suited for language tasks like writing and editing, OpenAI o1 is focused on multistep “reasoning,” the type of process required for advanced mathematics, coding, or other STEM-based questions. It uses a “chain of thought” technique, according to OpenAI. “It learns to recognize and correct its mistakes. It learns to break down tricky steps into simpler ones. It learns to try a different approach when the current one isn’t working,” the company wrote in a blog post on its website.

OpenAI’s tests point to resounding success. The model ranks in the 89th percentile on questions from the competitive coding organization Codeforces and would be among the top 500 high school students in the USA Math Olympiad, which covers geometry, number theory, and other math topics. The model is also trained to answer PhD-level questions in subjects ranging from astrophysics to organic chemistry. 

In math olympiad questions, the new model is 83.3% accurate, versus 13.4% for GPT-4o. In the PhD-level questions, it averaged 78% accuracy, compared with 69.7% from human experts and 56.1% from GPT-4o. (In light of these accomplishments, it’s unsurprising the new model was pretty good at writing a poem for our nuptial games, though still not perfect; it used more Ts and Ss than instructed to.)

So why does this matter? The bulk of LLM progress until now has been language-driven, resulting in chatbots or voice assistants that can interpret, analyze, and generate words. But in addition to getting lots of facts wrong, such LLMs have failed to demonstrate the types of skills required to solve important problems in fields like drug discovery, materials science, coding, or physics. OpenAI’s o1 is one of the first signs that LLMs might soon become genuinely helpful companions to human researchers in these fields. 

It’s a big deal because it brings “chain-of-thought” reasoning in an AI model to a mass audience, says Matt Welsh, an AI researcher and founder of the LLM startup Fixie. 

“The reasoning abilities are directly in the model, rather than one having to use separate tools to achieve similar results. My expectation is that it will raise the bar for what people expect AI models to be able to do,” Welsh says.

That said, it’s best to take OpenAI’s comparisons to “human-level skills” with a grain of salt, says Yves-Alexandre de Montjoye, an associate professor in math and computer science at Imperial College London. It’s very hard to meaningfully compare how LLMs and people go about tasks such as solving math problems from scratch.

Also, AI researchers say that measuring how well a model like o1 can “reason” is harder than it sounds. If it answers a given question correctly, is that because it successfully reasoned its way to the logical answer? Or was it aided by a sufficient starting point of knowledge built into the model? The model “still falls short when it comes to open-ended reasoning,” Google AI researcher François Chollet wrote on X.

Finally, there’s the price. This reasoning-heavy model doesn’t come cheap. Though access to some versions of the model is included in premium OpenAI subscriptions, developers using o1 through the API will pay three times as much as they pay for GPT-4o—$15 per 1 million input tokens in o1, versus $5 for GPT-4o. The new model also won’t be most users’ first pick for more language-heavy tasks, where GPT-4o continues to be the better option, according to OpenAI’s user surveys. 

What will it unlock? We won’t know until researchers and labs have the access, time, and budget to tinker with the new mode and find its limits. But it’s surely a sign that the race for models that can outreason humans has begun. 

Now read the rest of The Algorithm


Deeper learning

Chatbots can persuade people to stop believing in conspiracy theories

Researchers believe they’ve uncovered a new tool for combating false conspiracy theories: AI chatbots. Researchers from MIT Sloan and Cornell University found that chatting about a conspiracy theory with a large language model (LLM) reduced people’s belief in it by about 20%—even among participants who claimed that their beliefs were important to their identity. 

Why this matters: The findings could represent an important step forward in how we engage with and educate people who espouse such baseless theories, says Yunhao (Jerry) Zhang, a postdoc fellow affiliated with the Psychology of Technology Institute who studies AI’s impacts on society. “They show that with the help of large language models, we can—I wouldn’t say solve it, but we can at least mitigate this problem,” he says. “It points out a way to make society better.” Read more from Rhiannon Williams here.

Bits and bytes

Google’s new tool lets large language models fact-check their responses

Called DataGemma, it uses two methods to help LLMs check their responses against reliable data and cite their sources more transparently to users. (MIT Technology Review)

Meet the radio-obsessed civilian shaping Ukraine’s drone defense 

Since Russia’s invasion, Serhii “Flash” Beskrestnov has become an influential, if sometimes controversial, force—sharing expert advice and intel on the ever-evolving technology that’s taken over the skies. His work may determine the future of Ukraine, and wars far beyond it. (MIT Technology Review)

Tech companies have joined a White House commitment to prevent AI-generated sexual abuse imagery

The pledges, signed by firms like OpenAI, Anthropic, and Microsoft, aim to “curb the creation of image-based sexual abuse.” The companies promise to set limits on what models will generate and to remove nude images from training data sets where possible.  (Fortune)

OpenAI is now valued at $150 billion

The valuation arose out of talks it’s currently engaged in to raise $6.5 billion. Given that OpenAI is becoming increasingly costly to operate, and could lose as much as $5 billion this year, it’s tricky to see how it all adds up. (The Information)

There are more than 120 AI bills in Congress right now

More than 120 bills related to regulating artificial intelligence are currently floating around the US Congress.

They’re pretty varied. One aims to improve knowledge of AI in public schools, while another is pushing for model developers to disclose what copyrighted material they use in their training.  Three deal with mitigating AI robocalls, while two address biological risks from AI. There’s even a bill that prohibits AI from launching a nuke on its own.

The flood of bills is indicative of the desperation Congress feels to keep up with the rapid pace of technological improvements. “There is a sense of urgency. There’s a commitment to addressing this issue, because it is developing so quickly and because it is so crucial to our economy,” says Heather Vaughan, director of communications for the US House of Representatives Committee on Science, Space, and Technology.

Because of the way Congress works, the majority of these bills will never make it into law. But simply taking a look at all the different bills that are in motion can give us insight into policymakers’ current preoccupations: where they think the dangers are, what each party is focusing on, and more broadly, what vision the US is pursuing when it comes to AI and how it should be regulated.

That’s why, with help from the Brennan Center for Justice, which created a tracker with all the AI bills circulating in various committees in Congress right now, MIT Technology Review has taken a closer look to see if there’s anything we can learn from this legislative smorgasbord. 

As you can see, it can seem as if Congress is trying to do everything at once when it comes to AI. To get a better sense of what may actually pass, it’s useful to look at what bills are moving along to potentially become law. 

A bill typically needs to pass a committee, or a smaller body of Congress, before it is voted on by the whole Congress. Many will fall short at this stage, while others will simply be introduced and then never spoken of again. This happens because there are so many bills presented in each session, and not all of them are given equal consideration. If the leaders of a party don’t feel a bill from one of its members can pass, they may not even try to push it forward. And then, depending on the makeup of Congress, a bill’s sponsor usually needs to get some members of the opposite party to support it for it to pass. In the current polarized US political climate, that task can be herculean. 

Congress has passed legislation on artificial intelligence before. Back in 2020, the National AI Initiative Act was part of the Defense Authorization Act, which invested resources in AI research and provided support for public education and workforce training on AI.

And some of the current bills are making their way through the system. The Senate Commerce Committee pushed through five AI-related bills at the end of July. The bills focused on authorizing the newly formed US AI Safety Institute (AISI) to create test beds and voluntary guidelines for AI models. The other bills focused on expanding education on AI, establishing public computing resources for AI research, and criminalizing the publication of deepfake pornography. The next step would be to put the bills on the congressional calendar to be voted on, debated, or amended.

“The US AI Safety Institute, as a place to have consortium building and easy collaboration between corporate and civil society actors, is amazing. It’s exactly what we need,” says Yacine Jernite, an AI researcher at Hugging Face.

The progress of these bills is a positive development, says Varun Krovi, executive director of the Center for AI Safety Action Fund. “We need to codify the US AI Safety Institute into law if you want to maintain our leadership on the global stage when it comes to standards development,” he says. “And we need to make sure that we pass a bill that provides computing capacity required for startups, small businesses, and academia to pursue AI.”

Following the Senate’s lead, the House Committee on Science, Space, and Technology just passed nine more bills regarding AI on September 11. Those bills focused on improving education on AI in schools, directing the National Institute of Standards and Technology (NIST) to establish guidelines for artificial-intelligence systems, and expanding the workforce of AI experts. These bills were chosen because they have a narrower focus and thus might not get bogged down in big ideological battles on AI, says Vaughan.

“It was a day that culminated from a lot of work. We’ve had a lot of time to hear from members and stakeholders. We’ve had years of hearings and fact-finding briefings on artificial intelligence,” says Representative Haley Stevens, one of the Democratic members of the House committee.

Many of the bills specify that any guidance they propose for the industry is nonbinding and that the goal is to work with companies to ensure safe development rather than curtail innovation. 

For example, one of the bills from the House, the AI Development Practices Act, directs NIST to establish “voluntary guidance for practices and guidelines relating to the development … of AI systems” and a “voluntary risk management framework.” Another bill, the AI Advancement and Reliability Act, has similar language. It supports “the development of voluntary best practices and technical standards” for evaluating AI systems. 

“Each bill contributes to advancing AI in a safe, reliable, and trustworthy manner while fostering the technology’s growth and progress through innovation and vital R&D,” committee chairman Frank Lucas, an Oklahoma Republican, said in a press release on the bills coming out of the House.

“It’s emblematic of the approach that the US has taken when it comes to tech policy. We hope that we would move on from voluntary agreements to mandating them,” says Krovi.

Avoiding mandates is a practical matter for the House committee. “Republicans don’t go in for mandates for the most part. They generally aren’t going to go for that. So we would have a hard time getting support,” says Vaughan. “We’ve heard concerns about stifling innovation, and that’s not the approach that we want to take.” When MIT Technology Review asked about the origin of these concerns, they were attributed to unidentified “third parties.” 

And fears of slowing innovation don’t just come from the Republican side. “What’s most important to me is that the United States of America is establishing aggressive rules of the road on the international stage,” says Stevens. “It’s concerning to me that actors within the Chinese Communist Party could outpace us on these technological advancements.”

But these bills come at a time when big tech companies have ramped up lobbying efforts on AI. “Industry lobbyists are in an interesting predicament—their CEOs have said that they want more AI regulation, so it’s hard for them to visibly push to kill all AI regulation,” says David Evan Harris, who teaches courses on AI ethics at the University of California, Berkeley. “On the bills that they don’t blatantly try to kill, they instead try to make them meaningless by pushing to transform the language in the bills to make compliance optional and enforcement impossible.”

“A [voluntary commitment] is something that is also only accessible to the largest companies,” says Jernite at Hugging Face, claiming that sometimes the ambiguous nature of voluntary commitments allows big companies to set definitions for themselves. “If you have a voluntary commitment—that is, ‘We’re going to develop state-of-the-art watermarking technology’—you don’t know what state-of-the-art means. It doesn’t come with any of the concrete things that make regulation work.”

“We are in a very aggressive policy conversation about how to do this right, and how this carrot and stick is actually going to work,” says Stevens, indicating that Congress may ultimately draw red lines that AI companies must not cross.

There are other interesting insights to be gleaned from looking at the bills all together. Two-thirds of the AI bills are sponsored by Democrats. This isn’t too surprising, since some House Republicans have claimed to want no AI regulations, believing that guardrails will slow down progress.

The topics of the bills (as specified by Congress) are dominated by science, tech, and communications (28%), commerce (22%), updating government operations (18%), and national security (9%). Topics that don’t receive much attention include labor and employment (2%), environmental protection (1%), and civil rights, civil liberties, and minority issues (1%).

The lack of a focus on equity and minority issues came into view during the Senate markup session at the end of July. Senator Ted Cruz, a Republican, added an amendment that explicitly prohibits any action “to ensure inclusivity and equity in the creation, design, or development of the technology.” Cruz said regulatory action might slow US progress in AI, allowing the country to fall behind China.

On the House side, there was also a hesitation to work on bills dealing with biases in AI models. “None of our bills are addressing that. That’s one of the more ideological issues that we’re not moving forward on,” says Vaughan.

The lead Democrat on the House committee, Representative Zoe Lofgren, told MIT Technology Review, “It is surprising and disappointing if any of my Republican colleagues have made that comment about bias in AI systems. We shouldn’t tolerate discrimination that’s overt and intentional any more than we should tolerate discrimination that occurs because of bias in AI systems. I’m not really sure how anyone can argue against that.”

After publication, Vaughan clarified that “[Bias] is one of the bigger, more cross-cutting issues, unlike the narrow, practical bills we considered that week. But we do care about bias as an issue,” and she expects it to be addressed within an upcoming House Task Force report.

One issue that may rise above the partisan divide is deepfakes. The Defiance Act, one of several bills addressing them, is cosponsored by a Democratic senator, Amy Klobuchar, and a Republican senator, Josh Hawley. Deepfakes have already been abused in elections; for example, someone faked Joe Biden’s voice for a robocall to tell citizens not to vote. And the technology has been weaponized to victimize people by incorporating their images into pornography without their consent. 

“I certainly think that there is more bipartisan support for action on these issues than on many others,” says Daniel Weiner, director of the Brennan Center’s Elections & Government Program. “But it remains to be seen whether that’s going to win out against some of the more traditional ideological divisions that tend to arise around these issues.” 

Although none of the current slate of bills have resulted in laws yet, the task of regulating any new technology, and specifically advanced AI systems that no one entirely understands, is difficult. The fact that Congress is making any progress at all may be surprising in itself. 

“Congress is not sleeping on this by any stretch of the means,” says Stevens. “We are evaluating and asking the right questions and also working alongside our partners in the Biden-Harris administration to get us to the best place for the harnessing of artificial intelligence.”

Update: We added further comments from the Republican spokesperson.

AI-generated content doesn’t seem to have swayed recent European elections 

AI-generated falsehoods and deepfakes seem to have had no effect on election results in the UK, France, and the European Parliament this year, according to new research. 

Since the beginning of the generative-AI boom, there has been widespread fear that AI tools could boost bad actors’ ability to spread fake content with the potential to interfere with elections or even sway the results. Such worries were particularly heightened this year, when billions of people were expected to vote in over 70 countries. 

Those fears seem to have been unwarranted, says Sam Stockwell, the researcher at the Alan Turing Institute who conducted the study. He focused on three elections over a four-month period from May to August 2024, collecting data on public reports and news articles on AI misuse. Stockwell identified 16 cases of AI-enabled falsehoods or deepfakes that went viral during the UK general election and only 11 cases in the EU and French elections combined, none of which appeared to definitively sway the results. The fake AI content was created by both domestic actors and groups linked to hostile countries such as Russia. 

These findings are in line with recent warnings from experts that the focus on election interference is distracting us from deeper and longer-lasting threats to democracy.   

AI-generated content seems to have been ineffective as a disinformation tool in most European elections this year so far. This, Stockwell says, is because most of the people who were exposed to the disinformation already believed its underlying message (for example, that levels of immigration to their country are too high). Stockwell’s analysis showed that people who were actively engaging with these deepfake messages by resharing and amplifying them had some affiliation or previously expressed views that aligned with the content. So the material was more likely to strengthen preexisting views than to influence undecided voters. 

Tried-and-tested election interference tactics, such as flooding comment sections with bots and exploiting influencers to spread falsehoods, remained far more effective. Bad actors mostly used generative AI to rewrite news articles with their own spin or to create more online content for disinformation purposes. 

“AI is not really providing much of an advantage for now, as existing, simpler methods of creating false or misleading information continue to be prevalent,” says Felix Simon, a researcher at the Reuters Institute for Journalism, who was not involved in the research. 

However, it’s hard to draw firm conclusions about AI’s impact upon elections at this stage, says Samuel Woolley, a disinformation expert at the University of Pittsburgh. That’s in part because we don’t have enough data.

“There are less obvious, less trackable, downstream impacts related to uses of these tools that alter civic engagement,” he adds.

Stockwell agrees: Early evidence from these elections suggests that AI-generated content could be more effective for harassing politicians and sowing confusion than changing people’s opinions on a large scale. 

Politicians in the UK, such as former prime minister Rishi Sunak, were targeted by AI deepfakes that, for example, showed them promoting scams or admitting to financial corruption. Female candidates were also targeted with nonconsensual sexual deepfake content, intended to disparage and intimidate them. 

“There is, of course, a risk that in the long run, the more that political candidates are on the receiving end of online harassment, death threats, deepfake pornographic smears—that can have a real chilling effect on their willingness to, say, participate in future elections, but also obviously harm their well-being,” says Stockwell. 

Perhaps more worrying, Stockwell says, his research indicates that people are increasingly unable to discern the difference between authentic and AI-generated content in the election context. Politicians are also taking advantage of that. For example, political candidates in the European Parliament elections in France have shared AI-generated content amplifying anti-immigration narratives without disclosing that they’d been made with AI. 

“This covert engagement, combined with a lack of transparency, presents in my view a potentially greater risk to the integrity of political processes than the use of AI by the general population or so-called ‘bad actors,’” says Simon. 

Google is funding an AI-powered satellite constellation that will spot wildfires faster

Early next year, Google and its partners plan to launch the first in a series of satellites that together would provide close-up, frequently refreshed images of wildfires around the world, offering data that could help firefighters battle blazes more rapidly, effectively, and safely.

The online search giant’s nonprofit and research arms have collaborated with the Moore Foundation, the Environmental Defense Fund, the satellite company Muon Space, and others to deploy 52 satellites equipped with custom-developed sensors over the coming years. 

The FireSat satellites will be able to spot fires as small as 5 by 5 meters (16 by 16 feet) on any speck of the globe. Once the full constellation is in place, the system should be capable of updating those images about every 20 minutes, the group says.

Those capabilities together would mark a significant upgrade over what’s available from the satellites that currently provide data to fire agencies. Generally, they can provide either high-resolution images that aren’t updated rapidly enough to track fires closely or frequently refreshed images that are relatively low-resolution.

The Earth Fire Alliance collaboration will also leverage Google’s AI wildfire tools, which have been trained to detect early indications of wildfires and track their progression, to draw additional insights from the data.

The images and analysis will be provided free to fire agencies around the world, helping to improve understanding of where fires are, where they’re moving, and how hot they’re burning. The information could help agencies stamp out small fires before they turn into raging infernos, place limited firefighting resources where they’ll do the most good, and evacuate people along the safest paths.

“In the satellite image of the Earth, a lot of things can be mistaken for a fire: a glint, a hot roof, smoke from another fire,” says Chris Van Arsdale, climate and energy research lead at Google Research and chairman of the Earth Fire Alliance. “Detecting fires becomes a game of looking for needles in a world of haystacks. Solving this will enable first responders to act quickly and precisely when a fire is detected.”

Some details of FireSat were unveiled earlier this year. But the organizations involved will announce additional information about their plans today, including the news that Google.org, the company’s charitable arm, has provided $13 million to the program and that the inaugural launch is scheduled to occur next year. 

Reducing the fog of war

The news comes as large fires rage across millions of acres in the western US, putting people and property at risk. The blazes include the Line Fire in Southern California, the Shoe Fly Fire in central Oregon, and the Davis Fire south of Reno, Nevada.

Wildfires have become more frequent, extreme, and dangerous in recent decades. That, in part, is a consequence of climate change: Rising temperatures suck the moisture from trees, shrubs, and grasses. But fires increasingly contribute to global warming as well. A recent study found that the fires that scorched millions of acres across Canada last year pumped out 3 billion tons of carbon dioxide, four times the annual pollution produced by the airline industry.

GOOGLE

Humans have also increased fire risk by suppressing natural fires for decades, which has allowed fuel to build up in forests and grasslands, and by constructing communities on the edge of wilderness boundaries without appropriate rules, materials, and safeguards

Observers say that FireSat could play an important role in combating fires, both by enabling fire agencies to extinguish small ones before they grow into large ones and by informing effective strategies for battling them once they’re crossed that point.

“What these satellites will do is reduce the fog of war,” says Michael Wara, director of the climate and energy policy program at Stanford University’s Woods Institute for the Environment, who is focused on fire policy issues. “Like when a situation is really dynamic and very dangerous for firefighters and they’re trying to make decisions very quickly about whether to move in to defend structures or try to evacuate people.” 

(Wara serves on the advisory board of the Moore Foundation’s Wildfire Resilience Initiative.)

Some areas, like California, already have greater visibility into the current state of fires or early signs of outbreaks, thanks to technology like Department of Defense satellites, remote camera networks, and planes, helicopters, and drones. But FireSat will be especially helpful for “countries that have less-well-resourced wildland fighting capability,” Wara adds.

Better images, more data, and AI will not be able to fully counter the increased fire dangers. Wara and other fire experts argue that regions need to use prescribed burns and other efforts to more aggressively reduce the buildup of fuel, rethink where and how we build communities in fire-prone areas, and do more to fund and support the work of firefighters on the ground. 

Sounding an earlier alarm for fires will only help reduce dangers when regions have, or develop, the added firefighting resources needed to combat the most dangerous ones quickly and effectively. Communities will also need to put in place better policies to determine what types of fires should be left to burn, and under what conditions.

‘A game changer’

Kate Dargan Marquis, a senior wildfire advisor to the Moore Foundation who previously served as state fire marshal for California, says she can “personally attest” to the difference that such tools will make to firefighters in the field.

“It is a game changer, especially as wildfires are becoming more extreme, more frequent, and more dangerous for everyone,” she says. “Information like this will make a lifesaving difference for firefighters and communities around the globe.”

Kate Dargan Marquis, senior advisor, Moore Foundation.
GOOGLE

Google Research developed the sensors for the satellite and tested them as well as the company’s AI fire detection models by conducting flights over controlled burns in California. Google intends to work with Earth Fire Alliance “to ensure AI can help make this data as useful as possible, and also that wildfire information is shared as widely as possible,” the company said.

Google’s Van Arsdale says that providing visual images of every incident around the world from start to finish will be enormously valuable to scientists studying wildfires and climate change. 

“We can combine this data with Google’s existing models of the Earth to help advance our understanding of fire behavior and fire dynamics across all of Earth’s ecosystems,” he says. “All this together really has the potential to help mitigate the environmental and social impact of fire while also improving people’s health and safety.”

Specifically, it could improve assessments of fire risk, as well as our understanding of the most effective means of preventing or slowing the spread of fires. For instance, it could help communities determine where it would be most cost-effective to remove trees and underbrush. 

Figuring out the best ways to conduct such interventions is another key goal of the program, given their high cost and the limited funds available for managing wildlands, says Genny Biggs, the program director for the Moore Foundation’s Wildfire Resilience Initiative.

The launch

The idea for FireSat grew out of a series of meetings that began with a 2019 workshop hosted by the Moore Foundation, which provided the first philanthropic funding for the program. 

The first satellite, scheduled to be launched aboard a SpaceX rocket early next year, will be fully functional aside from some data transmission features. The goals of the “protoflight” mission include testing the onboard systems and the data they send back. The Earth Fire Alliance will work with a handful of early-adopter agencies to prepare for the next phases. 

The group intends to launch three fully operational satellites in 2026, with additional deployments in the years that follow. Muon Space will build and operate the satellites. 

Agencies around the world should be able to receive hourly wildfire updates once about half of the constellation is operational, says Brian Collins, executive director of the Earth Fire Alliance. It hopes to launch all 52 satellites by around the end of this decade.

Each satellite is designed to last about five years, so the organization will eventually need to deploy 10 more each year to maintain the constellation.

The Earth Fire Alliance has secured about two-thirds of the funding it needs for the first phase of the program, which includes the first four launches. The organization will need to raise additional money from government agencies, international organizations, philanthropies, and other groups  to deploy, maintain, and operate the full constellation. It estimates the total cost will exceed $400 million, which Collins notes “is 1/1000th of the economic losses due to extreme wildfires annually in the US alone.”

Asked if commercial uses of the data could also support the program, including potentially military ones, Collins said in an email: “Adjacent applications range from land use management and agriculture to risk management and industrial impact and mitigation.” 

“At the same time, we know that as large agencies and government agencies adopt FireSat data to support a broad public safety mandate, they may develop all-hazard, emergenc[y] management, and security related uses of data,” he added. “As long as opportunities are in balance with our charter to advance a global approach to wildfire and climate resilience, we welcome new ideas and applications of our data.”

‘Living with fire’

A wide variety of startups have emerged in recent years promising to use technology to reduce the frequency and severity of wildfires—for example, by installing cameras and sensors in forests and grasslands, developing robots to carry out controlled burns, deploying autonomous helicopters that can drop suppressant, and harnessing AI to predict wildfire behavior and inform forest and fire management strategies

So far, even with all these new tools, it’s still been difficult for communities to keep pace with the rising dangers.

Dargan Marquis—who founded her own wildfire software company, Intterra—says she is confident the incidence of disastrous fires can be meaningfully reduced with programs like FireSat, along with other improved technologies and policies. But she says it’s likely to take decades to catch up with the growing risks, as the world continues warming up.

“We’re going to struggle in places like California, these Mediterranean climates around the world, while our technology and our capabilities and our inventions, etc., catch up with that level of the problem,” she says. 

“We can turn that corner,” she adds. “If we work together on a comprehensive strategy with the right data and a convincing plan over the next 50 years, I do think that by the end of the century, we absolutely can be living with fire.”

Neuroscientists and architects are using this enormous laboratory to make buildings better

Have you ever found yourself lost in a building that felt impossible to navigate? Thoughtful building design should center on the people who will be using those buildings. But that’s no mean feat.

It’s not just about navigation, either. Just think of an office that left you feeling sleepy or unproductive, or perhaps a health center that had a less-than-reviving atmosphere. A design that works for some people might not work for others. People have different minds and bodies, and varying wants and needs. So how can we factor them all in?

To answer that question, neuroscientists and architects are joining forces at an enormous laboratory in East London—one that allows researchers to build simulated worlds. In this lab, scientists can control light, temperature, and sound. They can create the illusion of a foggy night, or the tinkle of morning birdsong.

And they can study how volunteers respond to these environments, whether they be simulations of grocery stores, hospitals, pedestrian crossings, or schools. That’s how I found myself wandering around a fake art gallery, wearing a modified baseball cap with a sensor that tracked my movements.

I first visited the Person-Environment-Activity Research Lab, referred to as PEARL, back in July. I’d been chatting to Hugo Spiers, a neuroscientist based at University College London, about the use of video games to study how people navigate. Spiers had told me he was working on another project: exploring how people navigate a lifelike environment, and how they respond during evacuations (which, depending on the situation, could be a matter of life or death).

For their research, Spiers and his colleagues set up what they call a “mocked-up art gallery” within PEARL. The center in its entirety is pretty huge as labs go, measuring around 100 meters in length and 40 meters across, with 10-meter-high ceilings in places. There’s no other research center in the world like this, Spiers told me.

The gallery setup looked a little like a maze from above, with a pathway created out of hanging black sheets. The exhibits themselves were videos of dramatic artworks that had been created by UCL students.

When I visited in July, Spiers and his colleagues were running a small pilot study to trial their setup. As a volunteer participant, I was handed a numbered black cap with a square board on top, marked with a large QR code. This code would be tracked by cameras above and around the gallery. The cap also carried a sensor, transmitting radio signals to devices around the maze that could pinpoint my location within a range of 15 centimeters.

At first, all the volunteers (most of whom seemed to be students) were asked to explore the gallery as we would any other. I meandered around, watching the videos, and eavesdropping on the other volunteers, who were chatting about their research and upcoming dissertation deadlines. It all felt pretty pleasant and calm.

That feeling dissipated in the second part of the experiment, when we were each given a list of numbers, told that each one referred to a numbered screen, and informed that we had to visit all the screens in the order in which they appeared on our lists. “Good luck, everybody,” Spiers said.

Suddenly everyone seemed to be rushing around, slipping past each other and trying to move quickly while avoiding collisions. “It’s all got a bit frantic, hasn’t it?” I heard one volunteer comment as I accidentally bumped into another. I hadn’t managed to complete the task by the time Spiers told us the experiment was over. As I walked to the exit, I noticed that some people were visibly out of breath.

The full study took place on Wednesday, September 11. This time, there were around 100 volunteers (I wasn’t one of them). And while almost everyone was wearing a modified baseball cap, some had more complicated gear, including EEG caps to measure brainwaves, or caps that use near-infrared spectroscopy to measure blood flow in the brain. Some people were even wearing eye-tracking devices that monitored which direction they were looking.

“We will do something quite remarkable today,” Spiers told the volunteers, staff, and observers as the experiment started. Taking such detailed measurements from so many individuals in such a setting represented “a world first,” he said.

I have to say that being an observer was much more fun than being a participant. Gone was the stress of remembering instructions and speeding around a maze. Here in my seat, I could watch as the data collected from the cameras and sensors was projected onto a screen. The volunteers, represented as squiggly colored lines, made their way through the gallery in a way that reminded me of the game Snake.

The study itself was similar to the pilot study, although this time the volunteers were given additional tasks. At one point, they were given an envelope with the name of a town or city in it, and asked to find others in the group who had been given the same one. It was fascinating to see the groups form. Some had the names of destination cities like Bangkok, while others had been assigned fairly nondescript English towns like Slough, made famous as the setting of the British television series The Office. At another point, the volunteers were asked to evacuate the gallery from the nearest exit.

The data collected in this study represents something of a treasure trove for researchers like Spiers and his colleagues. The team is hoping to learn more about how people navigate a space, and whether they move differently if they are alone or in a group. How do friends and strangers interact, and does this depend on whether they have certain types of material to bond over? How do people respond to evacuations—will they take the nearest exit as directed, or will they run on autopilot to the exit they used to enter the space in the first place?

All this information is valuable to neuroscientists like Spiers, but it’s also useful to architects like his colleague Fiona Zisch, who is based at UCL’s Bartlett School of Architecture. “We do really care about how people feel about the places we design for them,” Zisch tells me. The findings can guide not only the construction of new buildings, but also efforts to modify and redesign existing ones.

PEARL was built in 2021 and has already been used to help engineers, scientists, and architects explore how neurodivergent people use grocery stores, and the ideal lighting to use for pedestrian crossings, for example. Zisch herself is passionate about creating equitable spaces—particularly for health and education—that everyone can make use of in the best possible way.

In the past, models used in architecture have been developed with typically built, able-bodied men in mind. “But not everyone is a 6’2″ male with a briefcase,” Zisch tells me. Age, gender, height, and a range of physical and psychological factors can all influence how a person will use a building. “We want to improve not just the space, but the experience of the space,” says Zisch. Good architecture isn’t just about creating stunning features; it’s about subtle adaptations that might not even be noticeable to most people, she says.

The art gallery study is just the first step for researchers like Zisch and Spiers, who plan to explore other aspects of neuroscience and architecture in more simulated environments at PEARL. The team won’t have results for a while yet. But it’s a fascinating start. Watch this space.


Now read the rest of The Checkup

Read more from MIT Technology Review’s archive

Brain-monitoring technology has come a long way, and tech designed to read our minds and probe our memories is already being used. Futurist and legal ethicist Nita Farahany explained why we need laws to protect our cognitive liberty in a previous edition of The Checkup.

Listening in on the brain can reveal surprising insights into how this mysterious organ works. One team of neuroscientists found that our brains seem to oscillate between states of order and chaos.

Last year, MIT Technology Review published our design issue of the magazine. If you’re curious, this piece on the history and future of the word “design,” by Nicholas de Monchaux, head of architecture at MIT, might be a good place to start

Design covers much more than buildings, of course. Designers are creating new ways for users of prosthetic devices to feel more comfortable in their own skin—some of which have third thumbs, spikes, or “superhero skins.”

Achim Menges is an architect creating what he calls “self-shaping” structures with wood, which can twist and curve with changes in humidity. His approach is a low-energy way to make complex curved architectures, Menges told John Wiegand.

From around the web

Scientists are meant to destroy research samples of the poliovirus, as part of efforts to eradicate the disease it causes. But lab leaks of the virus may be more common than we’d like to think. (Science)

Neurofeedback allows people to watch their own brain activity in real time, and learn to control it. It could be a useful way to combat the impacts of stress. (Trends in Neurosciences)

Microbes, some of which cause disease in people, can travel over a thousand miles on wind, researchers have shown. Some appear to be able to survive their journey. (The Guardian)

Is the X chromosome involved in Alzheimer’s disease? A study of over a million people suggests so. (JAMA Neurology)

A growing number of men are paying thousands of dollars a year for testosterone therapies that are meant to improve their physical performance. But some are left with enlarged breasts, shrunken testicles, blood clots, and infertility. (The Wall Street Journal)

Meet the radio-obsessed civilian shaping Ukraine’s drone defense

Serhii “Flash” Beskrestnov hates going to the front line. The risks terrify him. “I’m really not happy to do it at all,” he says. But to perform his particular self-appointed role in the Russia-Ukraine war, he believes it’s critical to exchange the relative safety of his suburban home north of the capital for places where the prospect of death is much more immediate. “From Kyiv,” he says, “nobody sees the real situation.”

So about once a month, he drives hundreds of kilometers east in a homemade mobile intelligence center: a black VW van in which stacks of radio hardware connect to an array of antennas on the roof that stand like porcupine quills when in use. Two small devices on the dash monitor for nearby drones. Over several days at a time, Flash studies the skies for Russian radio transmissions and tries to learn about the problems facing troops in the fields and in the trenches.

He is, at least in an unofficial capacity, a spy. But unlike other spies, Flash does not keep his work secret. In fact, he shares the results of these missions with more than 127,000 followers—including many soldiers and government officials—on several public social media channels. Earlier this year, for instance, he described how he had recorded five different Russian reconnaissance drones in a single night—one of which was flying directly above his van.

“Brothers from the Armed Forces of Ukraine, I am trying to inspire you,” he posted on his Facebook page in February, encouraging Ukrainian soldiers to learn how to recognize enemy drone signals as he does. “You will spread your wings, you will understand over time how to understand distance and, at some point, you will save the lives of dozens of your colleagues.”

Drones have come to define the brutal conflict that has now dragged on for more than two and a half years. And most rely on radio communications—a technology that Flash has obsessed over since childhood. So while Flash is now a civilian, the former officer has still taken it upon himself to inform his country’s defense in all matters related to radio.

As well as the frontline information he shares on his public channels, he runs a “support service” for almost 2,000 military communications specialists on Signal and writes guides for building anti-drone equipment on a tight budget. “He’s a celebrity,” one special forces officer recently shouted to me over the thump of music in a Kyiv techno club. He’s “like a ray of sun,” an aviation specialist in Ukraine’s army told me. Flash tells me that he gets 500 messages every day asking for help.

Despite this reputation among rank-and-file service members—and maybe because of it—Flash has also become a source of some controversy among the upper echelons of Ukraine’s military, he tells me. The Armed Forces of Ukraine declined multiple requests for comment, but Flash and his colleagues claim that some high-ranking officials perceive him as a security threat, worrying that he shares too much information and doesn’t do enough to secure sensitive intel. As a result, some refuse to support or engage with him. Others, Flash says, pretend he doesn’t exist. Either way, he believes they are simply insecure about the value of their own contributions—“because everybody knows that Serhii Flash is not sitting in Kyiv like a colonel in the Ministry of Defense,” he tells me in the abrasive fashion that I’ve come to learn is typical of his character. 

But above all else, hours of conversations with numerous people involved in Ukraine’s defense, including frontline signalmen and volunteers, have made clear that even if Flash is a complicated figure, he’s undoubtedly an influential one. His work has become greatly important to those fighting on the ground, and he recently received formal recognition from the military for his contributions to the fight, with two medals of commendation—one from the commander of Ukraine’s ground forces, the other from the Ministry of Defense. 

With a handheld directional antenna and a spectrum analyzer, Flash can scan for hostile signals.
EMRE ÇAYLAK

Despite a small number of semi-autonomous machines with a reduced reliance on radio communications, the drones that saturate the skies above the battlefield will continue to largely depend on this technology for the foreseeable future. And in this race for survival—as each side constantly tries to best the other, only to start all over again when the other inevitably catches up—Ukrainian soldiers need to develop creative solutions, and fast. As Ukraine’s wartime radio guru, Flash may just be one of their best hopes for doing that. 

“I know nothing about his background,” says “Igrok,” who works with drones in Ukraine’s 110th Mechanized Brigade and whom we are identifying by his call sign, as is standard military practice. “But I do know that most engineers and all pilots know nothing about radios and antennas. His job is definitely one of the most powerful forces keeping Ukraine’s aerial defense in good condition.”

And given the mounting evidence that both militaries and militant groups in other parts of the world are now adopting drone tactics developed in Ukraine, it’s not only his country’s fate that Flash may help to determine—but also the ways that armies wage war for years to come.

A prescient hobby

Before I can even start asking questions during our meeting in May, Flash is rummaging around in the back of the Flash-mobile, pulling out bits of gear for his own version of show-and-tell: a drone monitor with a fin-shaped antenna; a walkie-talkie labeled with a sticker from Russia’s state security service, the FSB; an approximately 1.5-meter-long foldable antenna that he says probably came from a US-made Abrams tank.

Flash has parked on a small wooded road beside the Kyiv Sea, an enormous water reservoir north of the capital. He’s wearing a khaki sweat-wicking polo shirt, combat trousers, and combat boots, with a Glock 19 pistol strapped to his hip. (“I am a threat to the enemy,” he tells me, explaining that he feels he has to watch his back.) As we talk, he moves from one side to the other, as if the electromagnetic waves that he’s studied since childhood have somehow begun to control the motion of his body.

Now 49, Flash grew up in a suburb of Kyiv in the ’80s. His father, who was a colonel in the Soviet army, recalls bringing home broken radio equipment for his preteen son to tinker with. Flash showed talent from the start. He attended an after-school radio club, and his father fixed an antenna to the roof of their apartment for him. Later, Flash began communicating with people in countries beyond the Iron Curtain. “It was like an open door to the big world for me,” he says.

Flash recalls with amusement a time when a letter from the KGB arrived at his family home, giving his father the fright of his life. His father didn’t know that his son had sent a message on a prohibited radio frequency, and someone had noticed. Following the letter, when Flash reported to the service’s office in downtown Kyiv, his teenage appearance confounded them. Boy, what are you doing here? Flash recalls an embarrassed official saying. 

Ukraine had been a hub of innovation as part of the Soviet Union. But by the time Flash graduated from military communications college in 1997, Ukraine had been independent for six years, and corruption and a lack of investment had stripped away the armed forces’ former grandeur. Flash spent just a year working in a military radio factory before he joined a private communications company developing Ukraine’s first mobile network, where he worked with technologies far more advanced than what he had used in the military. The  project was called “Flash.” 

A decade and a half later, Flash had risen through the ranks of the industry to become head of department at the progenitor to the telecommunications company Vodafone Ukraine. But boredom prompted him to leave and become an entrepreneur. His many projects included a successful e-commerce site for construction services and a popular video game called Isotopium: Chernobyl, which he and a friend based on the “really neat concept,” according to a PC Gamer review, of allowing players to control real robots (fitted with radios, of course) around a physical arena. Released in 2019, it also received positive reviews from Reuters and BBC News.

But within just a few years, an unexpected attack would hurl his country into chaos—and upend Flash’s life. 

“I am here to help you with technical issues,” Flash remembers writing to his Signal group when he first started offering advice. “Ask me anything and I will try to find the answer for you.”
EMRE ÇAYLAK

By early 2022, rumors were growing of a potential attack from Russia. Though he was still working on Isotopium, Flash began to organize a radio network across the northern suburbs of Kyiv in preparation. Near his home, he set up a repeater about 65 meters above ground level that could receive and then rebroadcast transmissions from all the radios in its network across a 200-square-kilometer area. Another radio amateur programmed and distributed handheld radios.

When Russian forces did invade, on February 24, they took both fiber-optic and mobile networks offline, as Flash had anticipated. The radio network became the only means of instant communications for civilians and, critically, volunteers mobilizing to fight in the region, who used it to share information about Russian troop movements. Flash fed this intel to several professional Ukrainian army units, including a unit of special reconnaissance forces. He later received an award from the head of the district’s military administration for his part in Kyiv’s defense. The head of the district council referred to Flash as “one of the most worthy people” in the region.

Yet it was another of Flash’s projects that would earn him renown across Ukraine’s military.

Despite being more than 100 years old, radio technology is still critical in almost all aspects of modern warfare, from secure communications to satellite-guided missiles. But the decline of Ukraine’s military, coupled with the movement of many of the country’s young techies into lucrative careers in the growing software industry, created a vacuum of expertise. Flash leaped in to fill it.

Within roughly a month of Russia’s incursion, Flash had created a private group called “Military Signalmen” on the encrypted messaging platform Signal, and invited civilian radio experts from his personal network to join alongside military communications specialists. “I am here to help you with technical issues,” he remembers writing to the group. “Ask me anything and I will try to find the answer for you.”

The kinds of questions that Flash and his civilian colleagues answered in the first months were often basic. Group members wanted to know how to update the firmware on their devices, reset their radios’ passwords, or set up the internal communications networks for large vehicles. Many of the people drafted as communications specialists in the Ukrainian military had little relevant experience; Flash claims that even professional soldiers lacked appropriate training and has referred to large parts of Ukraine’s military communications courses as “either nonsense or junk.” (The Korolov Zhytomyr Military Institute, where many communications specialists train, declined a request for comment.)

After Russia’s invasion of Ukraine, Flash transformed his VW van into a mobile radio intelligence center.
EMRE ÇAYLAK

He demonstrates handheld spectrum analyzers with custom Ukrainian firmware.

News of the Signal group spread by word of mouth, and it soon became a kind of 24-hour support service that communications specialists in every sector of Ukraine’s frontline force subscribed to. “Any military engineer can ask anything and receive the answer within a couple of minutes,” Flash says. “It’s a nice way to teach people very quickly.” 

As the war progressed into its second year, Military Signalmen became, to an extent, self-sustaining. Its members had learned enough to answer one another’s questions themselves. And this is where several members tell me that Flash has contributed the most value. “The most important thing is that he brought together all these communications specialists in one team,” says Oleksandr “Moto,” a technician at an EU mission in Kyiv and an expert in Motorola equipment, who has advised members of the group. (He asked to not be identified by his surname, due to security concerns.) “It became very efficient.”

Today, Flash and his partners continue to answer occasional questions that require more advanced knowledge. But over the past year, as the group demanded less of his time, Flash has begun to focus on a rapidly proliferating weapon for which his experience had prepared him almost perfectly: the drone.  

A race without end

The Joker-10 drone, one of Russia’s latest additions to its arsenal, is equipped with a hibernation mechanism, Flash warned his Facebook followers in March. This feature allows the operator to fly it to a hidden location, leave it there undetected, and then awaken it when it’s time to attack. “It is impossible to detect the drone using radio-electronic means,” Flash wrote. “If you twist and turn it in your hands—it will explode.” 

This is just one example of the frequent developments in drone engineering that Ukrainian and Russian troops are adapting to every day. 

Larger strike drones similar to the US-made Reaper have been familiar in other recent conflicts, but sophisticated air defenses have rendered them less dominant in this war. Ukraine and Russia are developing and deploying vast numbers of other types of drones—including the now-notorious “FPV,” or first-person view, drone that pilots operate by wearing goggles that stream video of its perspective. These drones, which can carry payloads large enough to destroy tanks, are cheap (costing as little as $400), easy to produce, and difficult to shoot down. They use direct radio communications to transmit video feeds, receive commands, and navigate.

A Ukrainian soldier prepares an FPV drone equipped with dummy ammunition for a simulated flight operation.
MARCO CORDONE/SOPA IMAGES/SIPA USA VIA AP IMAGES

But their reliance on radio technology is a major vulnerability, because enemies can disrupt the signals that the drones emit—making them far less effective, if not inoperable. This form of electronic warfare—which most often involves emitting a more powerful signal at the same frequency as the operator’s—is called “jamming.”

Jamming, though, is an imperfect solution. Like drones, jammers themselves emit radio signals that can enable enemies to locate them. There are also effective countermeasures to bypass jammers. For example, a drone operator can use a tactic called “frequency hopping,” rapidly jumping between different frequencies to avoid a jammer’s signal. But even this method can be disrupted by algorithms that calculate the hopping patterns.

For this reason, jamming is a frequent focus of Flash’s work. In a January post on his Telegram channel, for instance, which people viewed 48,000 times, Flash explained how jammers used by some Ukrainian tanks were actually disrupting their own communications. “The cause of the problems is not direct interference with the reception range of the radio station, but very powerful signals from several [electronic warfare] antennae,” he wrote, suggesting that other tank crews experiencing the same problem might try spreading their antennas across the body of the tank. 

It is all part of an existential race in which Russia and Ukraine are constantly hunting for new methods of drone operation, drone jamming, and counter-jamming—and there’s no end in sight. In March, for example, Flash says, a frontline contact sent him photos of a Russian drone with what looks like a 10-kilometer-long spool of fiber-optic cable attached to its rear—one particularly novel method to bypass Ukrainian jammers. “It’s really crazy,” Flash says. “It looks really strange, but Russia showed us that this was possible.”

Flash’s trips to the front line make it easier for him to track developments like this. Not only does he monitor Russian drone activity from his souped-up VW, but he can study the problems that soldiers face in situ and nurture relationships with people who may later send him useful intel—or even enemy equipment they’ve seized. “The main problem is that our generals are located in Kyiv,” Flash says. “They send some messages to the military but do not understand how these military people are fighting on the front.”

Besides the advice he provides to Ukrainian troops, Flash also publishes online his own manuals for building and operating equipment that can offer protection from drones. Building their own tools can be soldiers’ best option, since Western military technology is typically expensive and domestic production is insufficient. Flash recommends buying most of the parts on AliExpress, the Chinese e-commerce platform, to reduce costs.

While all his activity suggests a close or at least cooperative relationship between Flash and Ukraine’s military, he sometimes finds himself on the outside looking in. In a post on Telegram in May, as well as during one of our meetings, Flash shared one of his greatest disappointments of the war: the military’s refusal of his proposal to create a database of all the radio frequencies used by Ukrainian forces. But when I mentioned this to an employee of a major electronic warfare company, who requested anonymity to speak about the sensitive subject, he suggested that the only reason Flash still complains about this is that the military hasn’t told him it already exists. (Given its sensitivity, MIT Technology Review was unable to independently confirm the existence of this database.) 

Flash believes that generals in Kyiv “do not understand how these military people are fighting on the front.” So even though he doesn’t like the risks they involve, he takes trips to the frontline about once a month.
EMRE ÇAYLAK

This anecdote is emblematic of Flash’s frustration with a military complex that may not always want his involvement. Ukraine’s armed forces, he has told me on several occasions, make no attempt to collaborate with him in an official manner. He claims not to receive any financial support, either. “I’m trying to help,” he says. “But nobody wants to help me.”

Both Flash and Yurii Pylypenko, another radio enthusiast who helps Flash manage his Telegram channel, say military officials have accused Flash of sharing too much information about Ukraine’s operations. Flash claims to verify every member of his closed Signal groups, which he says only discuss “technical issues” in any case. But he also admits the system is not perfect and that Russians could have gained access in the past. Several of the soldiers I interviewed for this story also claimed to have entered the groups without Flash’s verification process. 

It’s ultimately difficult to determine if some senior staff in the military hold Flash at arm’s length because of his regular, often strident criticism—or whether Flash’s criticism is the result of being held at arm’s length. But it seems unlikely either side’s grievances will subside soon; Pylypenko claims that senior officers have even tried to blackmail him over his involvement in Flash’s work. “They blame my help,” he wrote to me over Telegram, “because they think Serhii is a Russian agent reposting Russian propaganda.” 

Is the world prepared?

Flash’s greatest concern now is the prospect of Russia overwhelming Ukrainian forces with the cheap FPV drones. When they first started deploying FPVs, both sides were almost exclusively targeting expensive equipment. But as production has increased, they’re now using them to target individual soldiers, too. Because of Russia’s production superiority, this poses a serious danger—both physical and psychological—to Ukrainian soldiers. “Our army will be sitting under the ground because everybody who goes above ground will be killed,” Flash says. Some reports suggest that the prevalence of FPVs is already making it difficult for soldiers to expose themselves at all on the battlefield.

To combat this threat, Flash has a grand yet straightforward idea. He wants Ukraine to build a border “wall” of jamming systems that cover a broad range of the radio spectrum all along the front line. Russia has already done this itself with expensive vehicle-based systems, but these present easy targets for Ukrainian drones, which have destroyed several of them. Flash’s idea is to use a similar strategy, albeit with smaller, cheaper systems that are easier to replace. He claims, however, that military officials have shown no interest.

Although Flash is unwilling to divulge more details about this strategy (and who exactly he pitched it to), he believes that such a wall could provide a more sustainable means of protecting Ukrainian troops. Nevertheless, it’s difficult to say how long such a defense might last. Both sides are now in the process of developing artificial-intelligence programs that allow drones to lock on to targets while still outside enemy jamming range, rendering them jammer-proof when they come within it. Flash admits he is concerned—and he doesn’t appear to have a solution.

Flash admits he is worried about Russia overwhelming Ukrainian forces with the cheap FPV drones: “Our army will be sitting under the ground because everybody who goes above ground will be killed.”
EMRE ÇAYLAK

He’s not alone. The world is entirely unprepared for this new type of warfare, says Yaroslav Kalinin, a former Ukrainian intelligence officer and the CEO of Infozahyst, a manufacturer of equipment for electronic warfare. Kalinin recounts talking at an electronic-warfare-focused conference in Washington, DC, last December where representatives from some Western defense companies weren’t able to recognize the basic radio signals emitted by different types of drones. “Governments don’t count [drones] as a threat,” he says. “I need to run through the streets like a prophet—the end is near!”

Nevertheless, Ukraine has become, in essence, a laboratory for a new era of drone warfare—and, many argue, a new era of warfare entirely. Ukraine’s and Russia’s soldiers are its technicians. And Flash, who sometimes sleeps curled up in the back of his van while on the road, is one of its most passionate researchers. “Military developers from all over the world come to us for experience and advice,” he says. Only time will tell whether their contributions will be enough to see Ukraine through to the other side of this war. 

Charlie Metcalfe is a British journalist. He writes for magazines and newspapers, including Wired, the Guardian, and MIT Technology Review.

Meet 2024’s climate innovators under 35

This article is from The Spark, MIT Technology Review’s weekly climate newsletter. To receive it in your inbox every Wednesday, sign up here.

One way to know where a field is going? Take a look at what the sharpest new innovators are working on.

Good news for all of us: MIT Technology Review’s list of 35 Innovators Under 35 just dropped. And a decent number of the people who made the list are working in fields that touch climate and energy in one way or another.

Looking through, I noticed a few trends that might provide some hints about the future of climate tech. Let’s dig into this year’s list and consider what these innovators’ work might mean for efforts to combat climate change.

Power to the people

Perhaps unsurprisingly, quite a few innovators on this list are working on energy—and many of them have an interest in making energy consistently available where and when it’s needed. Wind and solar are getting cheap, but we need solutions for when the sun isn’t shining and the wind isn’t blowing.

Tim Latimer cofounded Fervo Energy, a geothermal company hoping to provide consistently available, carbon-free energy using Earth’s heat. You may be familiar with his work, since Fervo was on our list of 15 Climate Tech Companies to Watch in 2023.

Another energy-focused innovator on the list is Andrew Ponec of Antora Energy, a company working to build thermal energy storage systems. Basically, the company’s technology heats up blocks when cheap renewables are available, and then stores that heat and delivers it to industrial processes that need constant power. (You, the readers, named thermal energy storage the readers’ choice on this year’s 10 Breakthrough Technologies list.)

Rock stars

While new ways of generating electricity and storing energy can help cut our emissions in the future, other people are focused on how to clean up the greenhouse gases already in the atmosphere. At this point, removing carbon dioxide from the atmosphere is basically required for any scenario where we limit warming to 1.5 °C over preindustrial levels. A few of the new class of innovators are turning to rocks for help soaking up and locking away atmospheric carbon. 

Noah McQueen cofounded Heirloom Carbon Technologies, a carbon removal company. The technology works by tweaking the way minerals soak up carbon dioxide from the air (before releasing it under controlled conditions, so they can do it all again). The company has plans for facilities that could remove hundreds of thousands of tons of carbon dioxide each year. 

Another major area of research focuses on how we might store captured carbon dioxide. Claire Nelson is the cofounder of Cella Mineral Storage, a company working on storage methods to better trap carbon dioxide underground once it’s been mopped up.  

Material world

Finally, some of the most interesting work on our new list of innovators is in materials. Some people are finding new ones that could help us address our toughest problems, and others are trying to reinvent old ones to clean up their climate impacts.

Julia Carpenter found a way to make a foam-like material from metal. Its high surface area makes it a stellar heat sink, meaning it can help cool things down efficiently. It could be a huge help in data centers, where 40% of energy demand goes to cooling.

And I spoke with Cody Finke, cofounder and CEO of Brimstone, a company working on cleaner ways of making cement. Cement alone is responsible for nearly 7% of global greenhouse-gas emissions, and about half of those come from chemical reactions necessary to make it. Finke and Brimstone are working to wipe out the need for these reactions by using different starting materials to make this crucial infrastructural glue.

Addressing climate change is a sprawling challenge, but the researchers and founders on this list are tackling a few of the biggest issues I think about every day. 

Ensuring that we can power our grid, and all the industrial processes that we rely on for the stuff in our daily lives, is one of the most substantial remaining challenges. Removing carbon dioxide from the atmosphere in an efficient, cheap process could help limit future warming and buy us time to clean up the toughest sectors. And finding new materials, and new methods of producing old ones, could be a major key to unlocking new climate solutions. 

To read more about the folks I mentioned here and other innovators working in climate change and beyond, check out the full list.


Now read the rest of The Spark

Related reading

Fervo Energy (cofounded by 2024 innovator Tim Latimer) showed last year that its wells can be used like a giant underground battery.

A growing number of companies—including Antora Energy, whose CEO Andrew Ponec is a 2024 innovator—are working to bring thermal energy storage systems to heavy industry.

Cement is one of our toughest challenges, as Brimstone CEO and 2024 innovator Cody Finke will tell you. I wrote about Brimstone and other efforts to reinvent cement earlier this year.

A plant with yellow flowers

Another thing

We need a whole lot of metals to address climate change, from the copper in transmission lines to the nickel in lithium-ion batteries that power electric vehicles. Some researchers think plants might be able to help. 

Roughly 750 species of plants are so-called hyperaccumulators, meaning they naturally soak up and tolerate relatively high concentrations of metal. A new program is funding research into how we might use this trait to help source nickel, and potentially other metals, in the future. Read the full story here.

Keeping up with climate  

A hurricane that recently formed in the Gulf of Mexico is headed for Louisiana, ending an eerily quiet few weeks of the season. (Scientific American)

→ After forecasters predicted a particularly active season, the lull in hurricane activity was surprising. (New Scientist)

Rising sea levels are one of the symptoms of a changing climate, but nailing down exactly what “sea level” means is more complicated than you might think. We’ve gotten better at measuring sea level over the past few centuries, though. (New Yorker)

The US Department of Energy’s Loan Programs Office has nearly $400 million in lending authority. This year’s election could shift the focus of that office drastically, making it a bellwether of how the results could affect energy priorities. (Bloomberg)

What if fusion power ends up working, but it’s too expensive to play a significant role on the grid? Some modelers think the technology will remain expensive and could come too late to make a dent in emissions. (Heatmap)

Electric-vehicle sales are up overall, but some major automakers are backing away from goals on zero-emissions vehicles. Even though sales are increasing, uptake is slower than many thought it would be, contributing to the nervous energy in the industry. (Canary Media)

It’s a tough time to be in the business of next-generation batteries. The woes of three startups reveal that difficult times are here, likely for a while. (The Information)