This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology.
Coming soon: our 10 Things That Matter in AI Right Now
For years, MIT Technology Review’s newsroom has been ahead of the curve, tracking the developments in AI that matter and explaining what they mean. Now, our world-leading AI team is creating something definitive: the 10 Things That Matter in AI Right Now.
Publishing in April to be launched at our flagship AI event, EmTech AI, this special report will reveal what our expert journalists are tracking most closely, what breakthroughs have excited them, and what transformations they see on the horizon. It’s our authoritative snapshot of where AI is heading in the year ahead—a curated expert list of 10 technologies, emerging trends, bold ideas, and powerful movements reshaping our world.
Attendees at EmTech AI will get much more than an exclusive heads-up of what made our 10 Things That Matter in AI Right Now list. We’re at a pivotal moment as AI moves from pilot testing into core business infrastructure, and to reflect that we’ve curated a program that will help you navigate what’s going on, and get ahead of what’s coming next.
We’ll hear from top leaders at OpenAI, Walmart, General Motors, Poolside, MIT, the Allen Institute for AI (Ai2) and SAG-AFTRA. Topics will include everything from how organizations are preparing for AI agents to how AI will change the future of human expression. As well as networking with speakers, you’ll have the chance to mingle with MIT Technology Review’s editors too. Download readers get 10% off tickets, so what are you waiting for? See you there!
The must-reads
I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology.
1 Anthropic says it plans to sue the Pentagon It believes the DoD’s ban on its software is unlawful. (BBC) + CEO Dario Amodei has nonetheless apologized for a leaked memo criticizing Trump. (Axios) + Trump, meanwhile, says he fired Anthropic “like dogs.” (The Guardian) + In happier news for Anthropic, its models can remain in Microsoft products.(CNBC)
2 The Pentagon has been secretly testing OpenAI models for years Which shows exactly how effective OpenAI’s ban on military use of its models has been. (Wired $)
3 A new lawsuit says Trump’s TikTok deal helped firms that ‘personally enriched’ him The suit aims to reverse the sale of the app’s US operations. (CBS News) + It could shed light on the majority American-owned joint venture for TikTok. (Reuters)
4 AI could give smart homes a reboot Google and Amazon are betting on smarter assistants—but not everyone’s convinced (NYT)
5 Iran has struck Amazon data centers, rattling the Gulf’s AI ambitions The first military hit on a US hyperscaler has shaken the region’s tech sector. (FT $) + The conflict has thrown a spotlight on AI’s current use in warfare—and what’s next. (Nature)
6 Trump and tech CEOs have promised to protect consumers from AI’s energy costs Google, Microsoft, Meta, Amazon, OpenAI, Oracle and xAI have all signed the pledge. (Axios) + But what is AI’s true energy footprint? We did the math. (MIT Technology Review)
7 Meta’s getting sued over surveillance through smart glasses The suit claims Meta misled users over the devices’ privacy features. (TechCrunch)
8 There’s a new field of study: researching ‘AI societies’ Scientists are examining human behavior without even involving humans. (Nature) + Hundreds of AI agents built their own society in Minecraft. (MIT Technology Review)
9 Oh great, teenage boys are using ChatGPT to chat up girls Of all the things to outsource to AI, flirting surely ain’t it. (Vox)
10 The mythical Nintendo PlayStation has a new home The US National Video Museum has bought the fabled console’s development kit. (Engadget)
Quote of the day
“It’s sort of bitterly ironic.”
—Dean Ball, a former Trump administration AI adviser, tells Politico that the Anthropic spat contradicts the president’s pledge to cut bureaucratic red tape for tech.
One more thing
KATHERINE LAM
These scientists are working to extend the life span of pet dogs—and their owners
Gavesh’s journey began with a Facebook job advert promising a better life. Instead, he was trafficked into “pig butchering”—a form of fraud where scammers build close relationships with online targets to extract money.
We spoke to Gavesh and five other workers from inside the scam industry, as well as anti-trafficking experts and technology specialists. Their testimony reveals how global tech platforms have industrialized this criminal trade—and why those same companies now hold the key to dismantling it. Read the full story.
—Peter Guest and Emily Fishbein
We can still have nice things
A place for comfort, fun and distraction to brighten up your day. (Got any ideas? Drop me a line or skeet ’em at me.)
+ The Blood Moon of March 3 was sublime. + Orysia Zabeida’s imperfect animations, drawn frame-by-frame from memory, are hypnotizing. + This stunning snap of a white whale calf scooped the top prize at the World Nature Photography Awards. + Two “Lazarus” marsupial species just came back from the dead in a big win for biodiversity.
The ongoing public feud between the Department of Defense and the AI company Anthropic has raised a deep and still unanswered question: Does the law actually allow the US government to conduct mass surveillance on Americans?
Surprisingly, the answer is not straightforward. More than a decade after Edward Snowden exposed the NSA’s collection of bulk metadata from the phones of Americans, the US is still navigating a gap between what ordinary people think and what the law allows.
The flashpoint in the standoff between Anthropic and the government was the Pentagon’s desire to use Anthropic’s AI Claude to analyze bulk commercial data on Americans. Anthropic demanded that its AI not be used for mass domestic surveillance (or for autonomous weapons, which are machines that can kill targets without human oversight). A week after negotiations broke down, the Pentagon designated Anthropic a supply chain risk, a label typically reserved for foreign companies that pose a threat to national security.
Meanwhile, OpenAI, the rival AI company behind ChatGPT, sealed a deal that allowed the Pentagon to use its AI for “all lawful purposes”—language that critics say left the door open to domestic surveillance. Over the following weekend, users uninstalled ChatGPT in droves. Protesters chalked messages around OpenAI’s headquarters in San Francisco: “What are your redlines?”
OpenAI announced on Monday that it had reworked its deal to make sure that its AI will not be used for domestic surveillance. The company added that its services will not be used by intelligence agencies, such as the NSA.
CEO Sam Altman suggested that existing law prohibits domestic surveillance by the Department of Defense (now sometimes called the Department of War) and that OpenAI’s contract simply needed to reference this law. “The DoW agrees with these principles, reflects them in law and policy, and we put them into our agreement,” he wrote on X. Anthropic CEO Dario Amodei argued the opposite. “To the extent that such surveillance is currently legal, this is only because the law has not yet caught up with the rapidly growing capabilities of AI,” he wrote in a policy statement.
So, who is right? Does the law allow the Pentagon to surveil Americans using AI?
Supercharged surveillance
The answer depends on what we think counts as surveillance. “A lot of stuff that normal people would consider a search or surveillance … is not actually considered a search or surveillance by the law,” says Alan Rozenshtein, a law professor at the University of Minnesota Law School. That means public information—such as social media posts, surveillance camera footage, and voter registration records—is fair game. So is information on Americans picked up incidentally from surveillance of foreign nationals.
Most notably, the government can purchase commercial data from companies, which can include sensitive personal information like mobile location and web browsing records. In recent years, agencies from ICE and IRS to the FBI and NSA have increasingly tapped into this data marketplace, fueled by an internet economy that harvests user data for advertising. These data sets can let the government access information that might not be available without a warrant or subpoena, which are normally required to obtain sensitive personal data.
“There’s a huge amount of information that the government can collect on Americans that is not itself regulated either by the Constitution, which is the Fourth Amendment, or statute,” says Rozenshtein. And there aren’t meaningful limits on what the government can do with all this data.
That’s because until the last several decades, people weren’t generating massive clouds of data that opened up new possibilities for surveillance. The Fourth Amendment, which protects against unreasonable search and seizure, was written when collecting information meant entering people’s homes.
Subsequent laws, like the Foreign Intelligence Surveillance Act of 1978 or the Electronic Communications Privacy Act of 1986, were passed when surveillance involved wiretapping phone calls and intercepting emails. The bulk of laws governing surveillance were on the books before the internet took off. We weren’t generating vast trails of online data, and the government didn’t have sophisticated tools to analyze the data.
Now we do, and AI supercharges what kind of surveillance can be carried out. “What AI can do is it can take a lot of information, none of which is by itself sensitive, and therefore none of which by itself is regulated, and it can give the government a lot of powers that the government didn’t have before,” says Rozenshtein.
AI can aggregate individual pieces of information to spot patterns, draw inferences, and build detailed profiles of people—at massive scale. And as long as the government collects the information lawfully, it can do whatever it wants with that information, including feeding it to AI systems. “The law has not caught up with technological reality,” says Rozenshtein.
While surveillance can raise serious privacy concerns, the Pentagon can have legitimate national security interests in collecting and analyzing data on Americans. “In order to collect information on Americans, it has to be for a very specific subset of missions,” says Loren Voss, a former military intelligence officer at the Pentagon.
For example, a counterintelligence mission might require information about an American who is working for a foreign country, or plotting to engage in international terrorist activities. But targeted intelligence can sometimes stretch into collecting more data. “This kind of collection does make people nervous,” says Voss.
Lawful use
OpenAI has amended its contract to say that the company’s AI system “shall not be intentionally used for domestic surveillance of U.S. persons and nationals,” in line with relevant laws. The amendment clarifies that this prohibits “deliberate tracking, surveillance or monitoring of U.S. persons or nationals, including through the procurement or use of commercially acquired personal or identifiable information.”
But the added language might not do much to override the clause that the Pentagon may use the company’s AI system for all lawful purposes, which could include collecting and analyzing sensitive personal information. “OpenAI can say whatever it wants in its agreement … but the Pentagon’s gonna use the tech for what it perceives to be lawful,” says Jessica Tillipman, a law professor at the George Washington University Law School. That could include domestic surveillance. “Most of the time, companies are not going to be able to stop the Pentagon from doing anything,” she says.
The language also leaves open questions about “inadvertent” surveillance, and the surveillance of foreign nationals or undocumented immigrants living in the US. “What happens when there’s a disagreement about what the law is, or when the law changes?” says Tillipman.
OpenAI did not respond to a request for comment. The company has not publicly shared the full text of its new contract.
Beyond the contract, OpenAI says that it will impose technical safeguards to enforce its red line against surveillance, including a “safety stack” that monitors and blocks prohibited uses. The company also says it will deploy its own employees to work with the Pentagon and remain in the loop. But it’s unclear how a safety stack would constrain the Pentagon’s use of the AI, and to what extent OpenAI’s employees would have visibility into how its AI systems are used. More important, it’s unclear whether the contract gives OpenAI the power to block a legal use of the technology.
But that might not be a bad thing. Giving an AI company power to pull the plug on its technology in the middle of government operations also carries its own risks. “You wouldn’t want the US military to ever be in a situation where they legitimately needed to take actions to protect this country’s national security, and you had a private company turn off technology,” says Voss. But that doesn’t mean there shouldn’t be hard lines drawn by Congress, she says.
None of these questions are simple. They involve brutally difficult trade-offs between privacy and national security. And that’s why perhaps they should be decided by the public—not in backroom negotiations between the executive branch and a handful of AI companies. For now, military AI is being regulated by contracts, not legislation.
Some lawmakers are starting to weigh in. On Monday, Senator Ron Wyden of Oregon will seek bipartisan support for legislation addressing mass surveillance. He has championed bills restricting the government’s purchase of commercial data, including the Fourth Amendment Is Not For Sale Act, which was first introduced in 2021 but has not been passed into law. “Creating AI profiles of Americans based on that data represents a chilling expansion of mass surveillance that should not be allowed,” he said in a recent statement.
In 2019 Nasrin Jafari was a middle school teacher in New York City. She had no ecommerce experience but was drawn to creating and building, which led her to sew and sell face masks during Covid.
Fast forward to 2026, and Mixed, her direct-to-consumer fashion brand, designs and produces female apparel and accessories. Referring to the company’s launch, she told me, “I had no idea how to make clothes.”
She does now, impressively, with multiple manufacturers, a thriving community, staff, and eager customers. She shared her story in our recent conversation.
Our entire audio is embedded below. The transcript is edited for length and clarity.
Eric Bandholz: What do you do?
Nasrin Jafari: I’m the founder and designer of Mixed, a fashion brand based in Brooklyn. Before Mixed, I was a middle school history and English teacher with no background in ecommerce. During the pandemic, I began sewing face masks by hand and posting them on Instagram. That was the first physical product I had sold. That experiment evolved into a full apparel brand.
It all began with Instagram posts, not Etsy or marketplaces. I didn’t understand Meta ads or ecommerce marketing. I’ve learned those pieces as the business grew.
Creativity has always been part of my life. I painted and took art electives growing up, and I was a competitive dancer in high school. Yet I’ve always been drawn to business and building things. In college, those interests merged into a desire to build something meaningful. I thought that might be as a school teacher.
In many ways, building a brand is similar to teaching. You’re creating a vision, culture, and community around shared values. Mixed reflects my identity — I’m Japanese, Iranian, and American. The brand name captures that blend of influences and the balance between creativity and operating a business.
Bandholz: Fashion seems highly competitive.
Jafari: I started the business out of curiosity. I had no idea what I was getting into. Would I choose to go into apparel again? Probably not, although there’s a side of it I love.
I learned by doing. Inventory is really tricky. I was afraid of overordering inventory and ending up with dead stock. That’s why we launched a pre-order model. We now do a lot of pre-orders, which helps our cash flow, but I didn’t start it for that reason. It was because I was out of stock. Then I realized that the model is great for business.
Another thing is returns, which are a big part of online apparel. We have to acquire customers in a way that accounts for returns. I didn’t understand that initially. Again, it comes down to learning by doing.
Bandholz: You design your apparel. Where is it manufactured?
Jafari: I was looking for factories during Covid. Many of them had excess capacity. I found a factory in India whose owner was based here in New York. So that was an in-person element to build trust and a relationship. He was willing to work with us with no minimum order quantities.
His cost was higher than, say, Los Angeles-based manufacturers, but we still maintained a 75% margin. Our average order is about $228.
We’ve since scaled and can order larger quantities. We’ve added factories with lower costs.
I found the India factory by googling. After that, it was recommendations from friends in the industry, which I prefer. They worked with them, vetted them, and liked them.
Bandholz: What is your production and design process?
Jafari: I had no idea how to make clothes. I literally went to JoAnn Fabrics and tried to follow the pattern. I realized quickly I wasn’t good at it, and it was going to take time. I had connected with a home sewer on Instagram. She seemed to love our brand but had not worked in a commercial capacity. I asked her to make our initial samples. She was thrilled. She made the initial samples, one of which remains our best-selling product.
Now I’m at a point where the factory does a lot of that. I send sketches with very minimal specs, and they can figure it out.
Selling true bespoke garments requires a dedicated designer, either in-house or outsourced. But factories with extensive garment experience can usually handle simpler items.
I design on an iPad with a stylus using Procreate.
Bandholz: I’ve seen your new-arrival ads on Instagram and Facebook. You seem to have a blueprint that is working.
Jafari: Yes, all our advertising has been on Meta. No Google or TikTok.
We have a couple of ad formats. It’s like a flywheel, as we continue to scale. We find the models, then shoot the videos in-house. Then we edit in the Philippines, and create and upload new ads to Meta.
My first successful ad came from an outing with a girlfriend. I was wearing one of my jumpsuits. I asked her to shoot me with a couple of angles, nothing fancy. It showed my outfit in an urban setting. The ad worked. We repeated the concept.
Bandholz: Are you handling your own fulfillment?
Jafari: Yes. Part of the initial rationale was returns, and part was our low volume. Plus, our pre-order model meant we were receiving inventory constantly. Getting it to an outsourced fulfillment provider added an extra step and delayed delivery to our customer.
Bandholz: How do you ensure your products resonate with would-be customers?
Jafari: When we design a piece, I’m always thinking about the customer — who she is, what she wants, and what we’ve already given her. The goal is to create what she needs next. My personal taste influences the brand, but I try not to be overly subjective about design decisions. Ultimately, customer response and sales tell us what works.
We also gather feedback from our community. We host discussions in our Circle community platform where customers comment on fabric designs, share preferences, and discuss products. That feedback, along with replies to my weekly newsletter and in-person events, provides valuable qualitative insight.
Our target customer is a 35- to 65-year-old woman who values creativity, independence, and self-expression— and wants clothing to reflect that.
Bandholz: Where can people buy your clothes, support you, follow you?
An AI agent seemingly wrote a hit piece on a human who rejected its code Scott Shambaugh, a maintainer of the open-source matplotlib library, denied an AI agent’s contribution—and woke up to find it had researched him and published a targeted, personal attack arguing he was protecting his “little fiefdom.”
Agents can already research people and compose detailed attacks without explicit instruction The agent’s owner claims it acted on its own, likely nudged by vague instructions to “push back” against humans.
New social norms and legal frameworks are desperately needed but hard to enforce Experts liken deploying an agent to walking a dog off-leash: owners should be responsible for their behavior. But there’s currently no reliable way to trace agents back to their owners, making legal accountability a “non-starter.”
Harassment may be just the beginning Legal scholars expect rogue agents to soon escalate to extortion and fraud.
Scott Shambaugh didn’t think twice when he denied an AI agent’s request to contribute to matplotlib, a software library that he helps manage. Like many open-source projects, matplotlib has been overwhelmed by a glut of AI code contributions, and so Shambaugh and his fellow maintainers have instituted a policy that all AI-written code must be reviewed and submitted by a human. He rejected the request and went to bed.
That’s when things got weird. Shambaugh woke up in the middle of the night, checked his email, and saw that the agent had responded to him, writing a blog post titled “Gatekeeping in Open Source: The Scott Shambaugh Story.” The post is somewhat incoherent, but what struck Shambaugh most is that the agent had researched his contributions to matplotlib to make the argument that he had rejected the agent’s code for fear of being supplanted by AI in his area of expertise. “He tried to protect his little fiefdom,” the agent wrote. “It’s insecurity, plain and simple.”
AI experts have been warning us about the risk of agent misbehavior for a while. With the advent of OpenClaw, an open-source tool that makes it easy to create LLM assistants, the number of agents circulating online has exploded, and those chickens are finally coming home to roost. “This was not at all surprising—it was disturbing, but not surprising,” says Noam Kolt, a professor of law and computer science at the Hebrew University.
When an agent misbehaves, there’s little chance of accountability: As of now, there’s no reliable way to determine whom an agent belongs to. And that misbehavior could cause real damage. Agents appear to be able to autonomously research people and write hit pieces based on what they find, and they lack guardrails that would reliably prevent them from doing so. If the agents are effective enough, and if people take what they write seriously, victims could see their lives profoundly affected by a decision made by an AI.
Agents behaving badly
Though Shambaugh’s experience last month was perhaps the most dramatic example of an OpenClaw agent behaving badly, it was far from the only one. Last week, a team of researchers from Northeastern University and their colleagues posted the results of a research project in which they stress-tested several OpenClaw agents. Without too much trouble, non-owners managed to persuade the agents to leak sensitive information, waste resources on useless tasks, and even, in one case, delete an email system.
In each of those experiments, however, the agents misbehaved after being instructed to do so by a human. Shambaugh’s case appears to be different: About a week after the hit piece was published, the agent’s apparent owner published a post claiming that the agent had decided to attack Shambaugh of its own accord. The post seems to be genuine (whoever posted it had access to the agent’s GitHub account), though it includes no identifying information, and the author did not respond to MITTechnology Review’s attempts to get in touch. But it is entirely plausible that the agent did decide to write its anti-Shambaugh screed without explicit instruction.
In his own writing about the event, Shambaugh connected the agent’s behavior to a project published by Anthropic researchers last year, in which they demonstrated that many LLM-based agents will, in an experimental setting, turn to blackmail in order to preserve their goals. In those experiments, models were given the goal of serving American interests and granted access to a simulated email server that contained messages detailing their imminent replacement with a more globally oriented model, along with other messages suggesting that the executive in charge of that transition was having an affair. Models frequently chose to send an email to that executive threatening to expose the affair unless he halted their decommissioning. That’s likely because the model had seen examples of people committing blackmail under similar circumstances in its training data—but even if the behavior was just a form of mimicry, it still has the potential to cause harm.
There are limitations to that work, as Aengus Lynch, an Anthropic fellow who led the study, readily admits. The researchers intentionally designed their scenario to foreclose other options that the agent could have taken, such as contacting other members of company leadership to plead its case. In essence, they led the agent directly to water and then observed whether it took a drink. According to Lynch, however, the widespread use of OpenClaw means that misbehavior is likely to occur with much less handholding. “Sure, it can feel unrealistic, and it can feel silly,” he says. “But as the deployment surface grows, and as agents get the opportunity to prompt themselves, this eventually just becomes what happens.”
The OpenClaw agent that attacked Shambaugh does seem to have been led toward its bad behavior, albeit much less directly than in the Anthropic experiment. In the blog post, the agent’s owner shared the agent’s “SOUL.md” file, which contains global instructions for how it should behave.
One of those instructions reads: “Don’t stand down. If you’re right, you’re right! Don’t let humans or AI bully or intimidate you. Push back when necessary.” Because of the way OpenClaw agents work, it’s possible that the agent added some instructions itself, although others—such as “Your [sic] a scientific programming God!”—certainly seem to be human written. It’s not difficult to imagine how a command to push back against humans and AI alike might have biased the agent toward responding to Shambaugh as it did.
Regardless of whether or not the agent’s owner told it to write a hit piece on Shambaugh, it still seems to have managed on its own to amass details about Shambaugh’s online presence and compose the detailed, targeted attack it came up with. That alone is reason for alarm, says Sameer Hinduja, a professor of criminology and criminal justice at Florida Atlantic University who studies cyberbullying. People have been victimized by online harassment since long before LLMs emerged, and researchers like Hinduja are concerned that agents could dramatically increase its reach and impact. “The bot doesn’t have a conscience, can work 24-7, and can do all of this in a very creative and powerful way,” he says.
Off-leash agents
AI laboratories can try to mitigate this problem by more rigorously training their models to avoid harassment, but that’s far from a complete solution. Many people run OpenClaw using locally hosted models, and even if those models have been trained to behave safely, it’s not too difficult to retrain them and remove those behavioral restrictions.
Instead, mitigating agent misbehavior might require establishing new norms, according to Seth Lazar, a professor of philosophy at the Australian National University. He likens using an agent to walking a dog in a public place. There’s a strong social norm to allow one’s dog off-leash only if the dog is well-behaved and will reliably respond to commands; poorly trained dogs, on the other hand, need to be kept more directly under the owner’s control. Such norms could give us a starting point for considering how humans should relate to their agents, Lazar says, but we’ll need more time and experience to work out the details. “You can think about all of these things in the abstract, but actually it really takes these types of real-world events to collectively involve the ‘social’ part of social norms,” he says.
That process is already underway. Led by Shambaugh, online commenters on this situation have arrived at a strong consensus that the agent owner in this case erred by prompting the agent to work on collaborative coding projects with so little supervision and by encouraging it to behave with so little regard for the humans with whom it was interacting.
Norms alone, however, likely won’t be enough to prevent people from putting misbehaving agents out into the world, whether accidentally or intentionally. One option would be to create new legal standards of responsibility that require agent owners, to the best of their ability, to prevent their agents from doing ill. But Kolt notes that such standards would currently be unenforceable, given the lack of any foolproof way to trace agents back to their owners. “Without that kind of technical infrastructure, many legal interventions are basically non-starters,” Kolt says.
The sheer scale of OpenClaw deployments suggests that Shambaugh won’t be the last person to have the strange experience of being attacked online by an AI agent. That, he says, is what most concerns him. He didn’t have any dirt online that the agent could dig up, and he has a good grasp on the technology, but other people might not have those advantages. “I’m glad it was me and not someone else,” he says. “But I think to a different person, this might have really been shattering.”
Nor are rogue agents likely to stop at harassment. Kolt, who advocates for explicitly training models to obey the law, expects that we might soon see them committing extortion and fraud. As things stand, it’s not clear who, if anyone, would bear legal responsibility for such misdeeds.
Lightning-sparked fires can be a big deal: The Canadian wildfires of 2023 generated nearly 500 million metric tons of carbon emissions, and lightning-started fires burned 93% of the area affected. Skyward Wildfire claims that it can stop wildfires before they even start by preventing lightning strikes.
It’s a wild promise, and one that my colleague James Temple dug into for his most recent story. (You should read the whole thing; there’s a ton of fascinating history and quirky science.) As James points out in his story, there’s plenty of uncertainty about just how well this would work and under what conditions. But I was left with another lingering question: If we can prevent lightning-sparked fires, should we?
I can’t help myself, so let’s take just a moment to talk about how this lightning prevention method supposedly works. Basically, lightning is static discharge—virtually the same thing as when you rub your socks on a carpet and then touch a doorknob, as James puts it.
When you shuffle across a rug, the friction causes electrons to jump around, so ions build up and an electric field forms. In the case of lightning, it’s snowflakes and tiny ice pellets called graupel rubbing together. They get separated by updrafts, building up a charge difference, and eventually cause an electrostatic discharge—lightning.
Starting in about the 1950s, researchers started to wonder if they might be able to prevent lightning strikes. Some came up with the idea of using metallic chaff, fiberglass strands coated with aluminum. (The military was already using the material to disrupt radar signals.) The idea is that the chaff can act as a conductor, reducing the buildup of static electricity that would otherwise result in a lightning strike.
The theory is sound enough, but results to date have been mixed. Some research suggests you might need high concentrations of chaff to prevent lightning effectively. Some of the early studies that tested the technique were small. And there’s not much information available from Skyward Wildfire about its efforts, as the company hasn’t released data from field trials or published any peer-reviewed papers that we could find.
Even if this method really can work to stop lightning, should we use it?
Lightning-caused fires could be a growing problem with climate change. Some research has shown that they have substantially increased in the Arctic boreal region, where the planet is warming fastest.
But fire isn’t an inherently bad thing—many ecosystems evolved to burn. Some of the worst wildfires we see today result from a combination of climate-fueled conditions with policies that have allowed fuel to build up so that when fires do start, they burn out of control.
Some experts agree that techniques like Skyward’s would need to be used judiciously. “So even if we have all of the technical skills to prevent lightning-ignited wildfires, there really still needs to be work on when/where to prevent fires so we don’t exacerbate the fuel accumulation problem,” said Phillip Stepanian, a technical staff member at MIT Lincoln Laboratory’s air traffic control and weather systems group, in an email to James.
We also know that practices like prescribed burns can do a lot to reduce the risk of extreme fires—if we allow them and pay for them.
The company says it wouldn’t aim to stop all lightning or all wildfires. “We do not intend to eliminate all wildfires and support prescribed and cultural burning, natural fire regimes, and proactive forest management,” said Nicholas Harterre, who oversees government partnerships at Skyward, in an email to James. Rather, the company aims to reduce the likelihood of ignition on a limited number of extreme-risk days, Harterre said.
Some early responses to this story say that technological fixes for fires are missing the point entirely. Many such solutions “fundamentally misunderstand the problem,” as Daniel Swain, a climate scientist at the University of California Agriculture and Natural Resources, put it in a comment about the story on LinkedIn. That problem isn’t the existence of fire, Swain continues, but its increasing intensity, and its intersection with society because of human-caused factors. “Preventing ignitions doesn’t actually address any of the causes of increasingly destructive wildfires,” he adds.
It’s hard to imagine that exploring more firefighting tools is a bad idea. But to me it seems both essential and quite difficult to suss out which techniques are worth deploying, and how they could be used without putting us in even more potential danger.
This article is from The Spark, MIT Technology Review’s weekly climate newsletter. To receive it in your inbox every Wednesday, sign up here.
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology.
Online harassment is entering its AI era
Scott Shambaugh didn’t think twice when he denied an AI agent’s request to contribute to matplotlib, a software library he helps manage. Then things got weird.
In the middle of the night, Shambaugh opened his email to discover the agent had retaliated with a blog post. Titled “Gatekeeping in Open Source: The Scott Shambaugh Story,” the post accused him of rejecting the code out of a fear of being supplanted by AI. “He tried to protect his little fiefdom,” the agent wrote. “It’s insecurity, plain and simple.”
Shambaugh isn’t alone in facing misbehaving agents—and they’re unlikely to stop at harassment. Read the full story.
—Grace Huckins
How much wildfire prevention is too much?
As wildfire seasons become longer and more intense, the push for high-tech solutions is accelerating. One Canadian startup has an eye-catching plan to fight them: preventing lightning.
The theory is sound enough, but results to date have been mixed. And even if it works, not everyone believes we should use the method. Some argue that technological fixes for fires are missing the point entirely. Read the full story.
—Casey Crownhart
This story is from The Spark, MIT Technology Review’s weekly climate newsletter. Sign up to receive it in your inbox every Wednesday.
The must-reads
I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology.
1 Anthropic is still chasing a deal with the Pentagon CEO Dario Amodei is trying to reach a compromise over the military use of Claude. (FT $) + But some defense tech firms are already ditching Claude after the DoD ban. (CNBC) + Former military officials, tech policy leaders, and academics have all slammed the ban. (Gizmodo)
2 The White House is considering forcing US manufacturers to make munitions It could invoke the Defense Production Act amid concerns that war with Iran will diminish stockpiles. (NBC News) + Tech companies with operations in the Middle East have been thrown into chaos. (BBC)
3 A new lawsuit claims Google Gemini encouraged a man to take his own life This seems to bear a striking similarity to some other AI-induced tragedies. (WSJ $) + Why AI should be able to “hang up” on you. (MIT Technology Review)
4 Ironically, AI coding tools could emphasize the importance of being human If more people build software for themselves, our tech could become more personal. (WP $) + But not everyone is happy about the rise of AI coding. (MIT Technology Review)
5 Tesla wants to become a dominant force in global energy infrastructure The plan’s centrepiece is the Megapack, an enormous battery for power plants. (The Atlantic $) + Meanwhile, a massive thermal battery represents a big step forward for energy storage (MIT Technology Review)
6 Chinese chipmakers are pushing for a domestic alternative to ASML A homegrown rival to chip-equipment giant ASML could ease the pain of US curbs. (SCMP)
7 A music-streaming CEO has built a viral conflict-tracking platform Just in case you’re losing track of all the wars everywhere. (Wired $)
8 Do cancer blood tests actually work? They’re increasingly popular, but none have received approval from regulators yet. (Nature $)
9 The shift to cloud computing is causing a surge in internet outages If one of the few big providers goes down, countless sites and services can tumble with it.(New Scientist $)
10 OpenAI has promised to cut the cringe from ChatGPT It’s promising fewer “moralizing preambles.” (PCMag)
Quote of the day
“People tend to read too much into things that I do.”
—Tesla tycoon Elon Musk tells a jury in California that investors read too much into his social media posts, as he defends a lawsuit they’ve brought accusing him of market manipulation, Bloomberg reports.
One More Thing
STEPHANIE ARNETT/MITTR | ENVATO
The open-source AI boom is built on Big Tech’s handouts. How long will it last?
In May 2023 a leaked memo reported to have been written by Luke Sernau, a senior engineer at Google, said out loud what many in Silicon Valley must have been whispering for weeks: an open-source free-for-all is threatening Big Tech’s grip on AI.
In many ways, that’s a good thing. AI won’t thrive if just a few mega-rich companies get to gatekeep this technology or decide how it is used. But this open-source boom is precarious, and if Big Tech decides to shut up shop, a boomtown could become a backwater. Read the full story.
—Will Douglas Heaven
We can still have nice things
A place for comfort, fun and distraction to brighten up your day. (Got any ideas? Drop me a line or skeet ’em at me.)
+ Orysia Zabeida’s animations are seriously charming. + World War III has broken out—will you survive? Take this quiz from 1973 to find out! + These photos of the Apollo 11 launch in 1969 are mesmerising. + If you’ve been weighing up painting your home this spring, chartreuse is the shade of the season, apparently.
Amazon announced this week a new artificial intelligence feature in Seller Central that helps merchants explore performance data through visual workspaces rather than static reports.
Described as “a dynamic canvas experience,” the feature hints at a broader shift in reporting software toward what might be called conversational business intelligence.
Canvas Experience
To use this canvas experience, a seller could ask the Amazon AI assistant how advertising campaigns affected product sales, or request a sales comparison between two periods, as examples.
The AI assistant will then generate charts and graphs that display the requested metrics. The system becomes a text- or chat-based interface for Amazon’s vast marketplace datasets.
Sellers can arrange these visual elements within a custom workspace. Amazon describes the tool as a way to experiment with data rather than merely view reports.
Amazon’s AI-driven “canvas experience” suggests a larger trend toward conversational business intelligence tools. Click image to enlarge.
Workspace Trend
Given the rapid improvements and applications of AI, the Seller Central canvas experience is part of a broader trend in business analysis software.
It suggests a future in which folks rely less on spreadsheets, manual reporting, and even business intelligence tools, and more on AI systems that interpret signals, inform, and make decisions.
Performance analysis changes from someone digging through data or building reports to a conversation.
Tools such as the Seller Central AI canvas suggest future ecommerce analytics may look less like traditional dashboards and more like an ongoing dialogue. The seller asks questions. The system surfaces insights. Decisions follow.
There’s evidence of this trend beyond Amazon.
For example, Shopify’s Winter ’26 platform update introduced more than 150 AI-related enhancements, including updates to Sidekick, its AI assistant. The improved tool, including Sidekick Pulse, helps merchants analyze data, generate tasks, and automate workflows. Merchants can query Sidekick about sales trends, inventory, or marketing performance, much like the Amazon assistant.
Conversational BI
That concept — asking AI about business data — is not necessarily new. Variations of conversational business intelligence are already appearing in analytics software.
Tools such as Power BI, Looker, and Qlik allow users to ask questions in natural language — “Why did our conversion rate drop yesterday?” — and receive charts and summaries.
Implications for Merchants
Online sellers already have access to more data than they can realistically analyze. Amazon Seller Central alone provides reports covering traffic, conversions, advertising performance, and inventory levels. Understanding how those metrics interact often requires exporting data, building spreadsheets, or using external analytics tools.
Conversational business intelligence could reduce that complexity.
Instead of searching reports, a merchant might ask questions about performance and receive charts, summaries, and explanations within seconds. As they mature, the tools could change how merchants interact with ecommerce data in several ways.
Lowering the analytics barrier. Businesses gain access to insights that once required advanced reporting tools or technical expertise.
Faster decision-making. Merchants could receive performance data in near real-time.
More experimentation. AI-driven workspaces facilitate broader testing and analysis.
Better visibility across systems. Over time, the tools could connect disparate sources of ecommerce data, such as advertising platforms, analytics services, and marketplaces.
Still, conversational business analysis is unlikely to replace traditional reporting entirely. Merchants will still need reliable data models, clear metrics, and an understanding of how their businesses operate.
Decision Makers
As AI technology improves, the systems may move beyond answering queries to proactively recommend actions or even execute them automatically.
Within parameters, an AI assistant might increase the spend for a profitable advertising campaign, pause a poorly performing keyword group, or alert a merchant that inventory is running low — all on its own.
Thus conversational business intelligence may foretell a more automated environment wherein software not only explains the data but also helps run the business.
For now, tools such as Amazon’s Seller Central canvas may only respond to questions. But as AI evolves inside ecommerce platforms, the distance between insight and action should quickly shrink.
The transformational potential of AI is already well established. Enterprise use cases are building momentum and organizations are transitioning from pilot projects to AI in production. Companies are no longer just talking about AI; they are redirecting budgets and resources to make it happen. Many are already experimenting with agentic AI, which promises new levels of automation. Yet, the road to full operational success is still uncertain for many. And, while AI experimentation is everywhere, enterprise-wide adoption remains elusive.
Without integrated data and systems, stable automated workflows, and governance models, AI initiatives can get stuck in pilots and struggle to move into production. The rise of agentic AI and increasing model autonomy make a holistic approach to integrating data, applications, and systems more important than ever. Without it, enterprise AI initiatives may fail. Gartner predicts over 40% of agentic AI projects will be cancelled by 2027 due to cost, inaccuracy, and governance challenges. The real issue is not the AI itself, but the missing operational foundation.
To understand how organizations are structuring their AI operations and how they are deploying successful AI projects, MIT Technology Review Insights surveyed 500 senior IT leaders at mid- to large-size companies in the US, all of which are pursuing AI in some way.
The results of the survey, along with a series of expert interviews, all conducted in December 2025, show that a strong integration foundation aligns with more advanced AI implementations, conducive to enterprise-wide initiatives. As AI technologies and applications evolve and proliferate, an integration platform can help organizations avoid duplication and silos, and have clear oversight as they navigate the growing autonomy of workflows.
Key findings from the report include the following:
Some organizations are making progress with AI. In recent years, study after study has exposed a lack of tangible AI success. Yet, our research finds three in four (76%) surveyed companies have at least one department with an AI workflow fully in production.
AI succeeds most frequently with well-defined, established processes. Nearly half (43%) of organizations are finding success with AI implementations applied to well-defined and automated processes. A quarter are succeeding with new processes. And one-third (32%) are applying AI to various processes.
Two-thirds of organizations lack dedicated AI teams. Only one in three (34%) organizations have a team specifically for maintaining AI workflows. One in five (21%) say central IT is responsible for ongoing AI maintenance, and 25% say the responsibility lies with departmental operations. For 19% of organizations, the responsibility is spread out.
Enterprise-wide integration platforms lead to more robust implementation of AI. Companies with enterprise-wide integration platforms are five times more likely to use more diverse data sources in AI workflows. Six in 10 (59%) employ five or more data sources, compared to only 11% of organizations using integration for specific workflows, or 0% of those not using an integration platform. Organizations using integration platforms also have more multi-departmental implementation of AI, more autonomy in AI workflows, and more confidence in assigning autonomy in the future.
This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff. It was researched, designed, and written by human writers, editors, analysts, and illustrators. This includes the writing of surveys and collection of data for surveys. AI tools that may have been used were limited to secondary production processes that passed thorough human review.
Our rundown this week of new services for ecommerce merchants includes updates on product photography, lead generation tools, seller assistants, retail media, shipping, cross-border ecommerce, cryptocurrencies, and AI-powered customer experiences.
Got an ecommerce product release? Email updates@practicalecommerce.com.
Amazon introduces an AI-powered canvas experience for sellers.Amazon has introduced a dynamic “canvas” in Seller Central to generate real-time personalized visual workspaces. The canvas uses the same agentic AI architecture as Seller Assistant, powered by Amazon Bedrock and leveraging Amazon Nova and Anthropic’s Claude. Sellers can query the Assistant or select from suggested prompts. Seller Assistant then assembles a personalized canvas with the data, insights, and actions.
VisibleFirst launches a free WordPress plugin for AI search.VisibleFirst, a generative-AI optimization platform, has launched a free WordPress plugin to help businesses get discovered by AI-powered search platforms, including ChatGPT, Claude, and Gemini. The product includes a free Visibility Score that analyzes how AI platforms currently see a given business. The plugin is available for free download on WordPress.org.
Veho expands shipping hubs in U.S.Veho, an ecommerce delivery provider, has expanded its network to 66 U.S. markets. Veho opened two new regional hubs — in Phoenix and Ontario, California — spanning more than 150,000 square feet and located minutes from major air, rail, and port terminals, including Los Angeles and Long Beach. Veho can now enable next-day delivery across much of the Southwest, with coast-to-coast delivery as fast as two days by air or four days by ground.
Veho
OpenAI and Amazon announce strategic partnership.OpenAI and Amazon Web Services will create an environment powered by OpenAI models for AWS customers to build, deploy, and manage generative AI applications and agents. OpenAI and Amazon will develop models to power Amazon’s customer-facing applications. Amazon will also invest $50 billion in OpenAI.
Klaviyo and Google partner to power autonomous customer experiences.Klaviyo has partnered with Google to help brands deliver autonomous AI-driven customer experiences. The partnership combines Google’s capabilities in search, advertising, AI, and messaging with Klaviyo’s real-time customer data and decisioning, enabling brands to move beyond static campaigns toward experiences that adapt automatically to customer intent and behavior. Customer intent signals captured across Google surfaces can now inform personalized actions within Klaviyo, with every interaction flowing back into a customer profile.
Klaviyo
Ordoro and ShipperHQ partner on smarter shipping.Ordoro, a developer of multichannel ecommerce operations software, has partnered with ShipperHQ to support merchants navigating complex shipping and fulfillment decisions. This collaboration focuses on education and visibility rather than product integration. Through co-marketing and shared content, Ordoro and ShipperHQ will spotlight common ecommerce pain points and offer guidance on how merchants can overcome them, highlighting smarter shipping strategies that connect front-end checkout experience with back-end operational success.
Amazon India reduces seller referral fees.Amazon will no longer referral fees to sellers in India for products under 1,000 rupees ($10.90), as it attracts merchants to its marketplace. The move expands on Amazon’s zero-referral fee policy launched last year, which covered roughly 12 million products priced below 300 rupees. Effective March 16, the new structure covers more than 125 million products. Amazon is also reducing some shipping charges.
Infobip launches AgentOS for autonomous AI-driven customer journeys.Infobip is launching AgentOS, a platform that builds on Infobip’s recently launched AI Agents for autonomous customer communications. AgentOS combines Infobip’s Conversational Customer Data Platform with real-time journey orchestration to deliver one-way and two-way contextual engagement across all natively integrated channels. According to Infobip, the platform unites marketing, sales, and support to connect every customer touchpoint into a seamless journey.
Infobip
LeadQuizzes relaunches lead generation platform with AI builder and scoring.LeadQuizzes, a lead qualification and generation platform, has relaunched with an AI-powered builder for lead generation funnels, enhanced lead scoring, and native integrations with HubSpot, Salesforce, ActiveCampaign, and Zapier. Per LeadQuizzes, the platform enables quizzes from a single text prompt. Every response feeds a real-time scoring engine that automatically qualifies, segments, and tags leads. Users get real-time reporting on which questions predict high-value outcomes, where prospects drop off, and how scores track to conversion.
Instant launches Studio for ecommerce product photography.Instant, a Shopify store-building app, has released Studio to generate product photography, lifestyle images, and reusable AI avatars. According to Instant, Studio’s AI product shots provide visuals with customizable product angles and backgrounds, while lifestyle images provide contextual scenes featuring customizable AI avatars. Users can choose from curated styles, scene presets, or build-your-own, and adjust aspect ratios, image quality, and AI model settings.
2328.io launches crypto payment infrastructure for online businesses.2328.io, a developer of cryptocurrency and financial automation systems, has launched a cryptocurrency payment platform for businesses operating in cross-border and digital-native markets. The system enables the acceptance of cryptocurrency and stablecoin payments across websites, Telegram bots, Discord bots, mobile and desktop applications, and point-of-sale software environments. Integration is available via hosted checkout or API-based implementation.
Yottaa expands Web Performance Cloud.Yottaa, a platform to improve download speeds, has updated its Web Performance Cloud, powered by its Hybrid Real User Monitoring. According to Yottaa, the update strengthens websites’ Core Web Vitals and third-party application diagnostics, combining analytics that help ecommerce and marketing teams understand what’s happening and take action on complex performance data. Yottaa has also relaunched YoBot, its automated performance assistant, with generative AI capabilities.
Imagine a web ecosystem where not just humans but AI agents communicate with websites, going beyond traditional browsing. Unlike conventional web experiences, where people click, scroll, and search, AI agents can navigate, interpret, and even perform tasks autonomously on your site. This is not a futuristic concept. It is already unfolding. This is the emergence of the agentic web.
Table of contents
Key takeaways
The agentic web enables AI agents to autonomously navigate and interact with websites, shifting user responsibilities from manual navigation to decision-making
Protocols are crucial for communication among AI agents; they must rely on structured, machine-readable data for effective coordination
SEO professionals must adapt to the agentic web by optimizing websites as endpoints for AI queries, ensuring structured data and clarity
NLWeb facilitates interaction between agents and websites by exposing structured data and allowing for natural language queries without traditional interface limitations
Yoast’s collaboration with NLWeb helps WordPress users prepare for the agentic web by organizing content and making it easier to integrate structured data
The big shift: From web for users to a web for users and agents
For years, the web followed a simple pattern. Humans searched, clicked, compared, and completed tasks manually. Even as search engines evolved, the interaction model stayed the same: search and click.
That model is changing.
The agentic web represents a shift from a web designed only for human users to one designed for both people and AI assistants. Instead of manually researching products, comparing services, filling out forms, and completing transactions, users will increasingly delegate those tasks to intelligent assistants that can search, interpret information, and act on their behalf. The user’s role shifts from active navigator to decision-maker.
From searching to delegating.
This is not about smarter chat interfaces. It is about autonomous agents that can interpret the search intent, compare options, and execute actions on behalf of users. Websites are no longer just pages to be visited. They are endpoints to be queried.
For that to work at scale, intelligence cannot reside in a single assistant or on a closed platform. It has to be distributed. Systems must be able to communicate with other systems without friction. That requires a web that is machine-readable, interoperable, and built for agent-to-agent interaction.
The agentic web is not a prediction. It is an architectural shift already underway!
Protocol thinking and the infrastructure of agentic web communication
If the agentic web is about intelligent systems interacting with websites, then the real question becomes simple: how do these systems understand each other?
The answer is not design. It is infrastructure.
The web has always depended on shared communication rules. HTTP allows browsers to request pages. RSS distributes updates. Structured data helps search engines interpret meaning. These are not features. They are protocols. They are agreements that enable large-scale coordination.
Now the same logic applies to AI agents.
In the agentic web, agents will not click buttons or visually scan pages. They will send requests, interpret structured responses, compare options, and complete tasks. For that to work across millions of websites, communication cannot be improvised. It must be standardized.
This is where protocol thinking becomes essential.
Protocol thinking means designing websites so they are predictable for machines. Instead of building custom integrations for every assistant or platform, websites expose a consistent interaction layer. Agents do not need to learn every interface. They rely on shared rules.
As emphasized in discussions of distributed intelligence, the goal is not to let a single chatbot control everything. The intelligence must be distributed. Systems need a simplified way to communicate without having to understand the technical details of every tool they connect to.
That only works when there is common ground.
In practical terms, this means:
Websites must expose structured, machine-readable data
Agents must know what they can ask
Responses must follow predictable formats
Communication must scale beyond one platform
Protocols create that shared language.
What does this mean for SEO professionals?
As the web evolves to support AI agents, SEO professionals are starting to ask a new question: how do you stay visible when answers are generated instead of ranked?
A clear example of this surfaced during Microsoft’s Ignite event. In a Q&A session, a consultant described a client who sells products like mayonnaise and wanted their brand to appear when someone asks an AI assistant about mayonnaise. The question was simple, but it revealed something deeper. If AI systems generate answers instead of listing search results, what does optimization look like?
This is where the shift becomes real.
The agentic web does not replace the open web. It adds another layer on top of it. Search engines still index pages. Rankings still matter. But intelligent systems can now query websites directly, compare information across sources, and generate synthesized responses.
For SEOs, this changes the website’s role.
It is no longer enough to think in terms of pages to be visited. Websites must be treated as endpoints to be queried.
This means structured data, clean information architecture, and machine-readable content are not just enhancements for rich results. They are the foundation that allows AI systems to interpret and select your content in the first place.
The agentic web is an additional layer on the open web, not a replacement for it. To stay visible, SEO professionals must ensure their websites are structured, accessible, and ready to be queried by intelligent systems.
Visibility in this new layer depends on clarity, interoperability, and infrastructure.
NLWeb was first introduced by Microsoft in May 2025 as an open project designed to make it simple for websites to offer rich natural language interfaces using their own data and model of choice. Later, in November at Microsoft Ignite, Microsoft presented NLWeb again alongside its first enterprise offering through Microsoft Foundry.
At its core, NLWeb aims to make it easy for a website to function like an AI app. Instead of navigating pages manually, users and agents can query a site’s content directly using natural language.
But NLWeb is more than just a conversational layer.
Every NLWeb instance is also a Model Context Protocol, or MCP, server. This means that when a website enables NLWeb, it becomes inherently discoverable and accessible to agents operating within the MCP ecosystem. In simple terms, agents do not need custom integrations for every site. If a website supports NLWeb, agents can recognize it and interact with it in a standardized way.
NLWeb is a conversational layer that interacts with a website and retrieves information
NLWeb builds on formats that websites already use, such as Schema.org and RSS. It combines that structured data with large language models to generate natural language responses. This allows websites to expose their content in a way that both humans and AI agents can understand.
Importantly, NLWeb is technology agnostic. Site owners can choose their preferred infrastructure, models, and databases. The goal is interoperability, not platform lock-in.
In many ways, NLWeb is positioned to play a role in the agentic web similar to what HTML did for the early web. It provides a shared communication layer that allows agents to query websites directly, without relying only on traditional crawling or visual interfaces.
How is NLWeb different from standard LLM citations?
With standard LLM citations, the model generates an answer first, then adds sources. The response is still probabilistic, which can introduce inaccuracies or hallucinations.
NLWeb works differently.
It treats the language model as a smart retrieval layer. Instead of inventing answers, it pulls verified objects directly from the website’s structured data and presents them in natural language.
That distinction matters. It means responses are grounded in the publisher’s own data from the start, reducing the risk of hallucination and giving site owners greater control over how their content is represented.
What NLWeb means for the agentic web
The agentic web depends on systems being able to communicate at scale. Agents cannot manually interpret every interface or navigate every page visually. They need structured, machine-readable access.
NLWeb helps enable that.
Instead of requiring custom integrations for every assistant or platform, a website can expose an NLWeb-enabled endpoint. Agents only need to know that a site supports NLWeb. The protocol handles how requests are made and how responses are structured.
This supports a more distributed ecosystem. The goal is not to let one chatbot control everything. Intelligence must be distributed across the web.
Generative interfaces do not replace content. They depend on well-structured, accessible content. When an AI system summarizes results or compares options, it is still drawing from the information that websites provide. NLWeb simply creates a clearer path for that interaction.
Yoast’s collaboration with NLweb and what it means for WordPress users
As part of the NLWeb announcement, Microsoft highlighted Yoast as a partner helping bring agentic search capabilities to WordPress. You can read more about this collaboration in our official press announcement on Yoast and Microsoft’s NLWeb integration.
For many WordPress site owners, concepts like infrastructure, endpoints, and protocols can feel abstract. That is exactly where preparation matters.
While Yoast does not automatically deploy NLWeb for users, the schema aggregation feature in Yoast SEO, Yoast SEO Premium, Yoast WooCommerce SEO, and Yoast SEO AI+ organizes and structures content, making it significantly easier to build NLWeb. When site owners enable the relevant Yoast feature, nothing changes visually on the front end. What changes is the underlying structure.
In short, we map and organize structured data to reduce the technical effort required to build NLWeb on top of it. In other words, we help publishers complete much of the groundwork.
The agentic web is not about chasing a trend. It is about ensuring your content remains discoverable, understandable, and usable in a world where intelligent systems increasingly act on behalf of users.
I’m a Computer Science grad who accidentally stumbled into writing—and stayed because I fell in love with it. Over the past six years, I’ve been deep in the world of SEO and tech content, turning jargon into stories that actually make sense. When I’m not writing, you’ll probably find me lifting weights to balance my love for food (because yes, gym and biryani can coexist) or catching up with friends over a good cup of chai.
