A new prenatal roadmap could soon help clinicians place endotracheal tubes (ETTs) more precisely in the tiniest preterm newborns. In infants with birth weights under 500 g, even small positioning errors during intubation can increase the risk of life‑threatening complications, including inadequate ventilation or airway trauma. Researchers say their approach uses measurements taken before birth—capturing the baby’s anatomy while it is still developing in utero.
The study focuses on predicting the optimal ETT depth rather than relying solely on standard postnatal formulas. Using prenatal biometry, investigators model how fetal growth and structural proportions relate to the distance from the mouth to the trachea. The goal is to translate ultrasound-derived anatomical signals into a practical estimate that can guide clinicians at the bedside.
Technically, the model links prenatal measurements to the anticipated airway dimensions at delivery. This requires defining which biometric inputs best correlate with the final tube placement target, then training a predictive framework to output a recommended insertion depth for extremely low birth weight infants. By emphasizing prenatal inputs, the method aims to reduce uncertainty that can arise from day-to-day postnatal changes.
Importantly, the research targets a subgroup with particularly high vulnerability. Infants below 500 g are at increased risk because their airway structures are extremely small and intubation margins are narrow. In this context, individualized prediction could improve consistency compared with one-size-fits-all guidelines.
If validated in larger clinical settings, the prenatal model could be integrated into obstetric and neonatal workflows. After routine prenatal ultrasound assessments, clinicians would generate an ETT depth estimate ahead of delivery, supporting more confident planning for immediate respiratory care.
The findings also highlight a broader trend in perinatal medicine: using quantitative prenatal data to optimize invasive procedures. Instead of reacting after intubation placement, the strategy moves decision-making earlier—toward anticipation, personalization, and safety.
Overall, the work suggests that prenatal biometry can carry actionable information for airway management. The team frames the approach as a step toward more precise, data-driven neonatal resuscitation and ventilation strategies for the smallest patients.
The study, published in Journal of Perinatology, is available under DOI: 10.1038/s41372-026-02851-2.
Subject of Research: Prenatal biometry-based prediction of optimal ETT depth Article Title: Prenatal biometry-based prediction of optimal endotracheal tube depth in infants with a birth weight less than 500 g. Article References: Yoo, K., Kim, S.H., Park, J. et al.J Perinatol (2026). https://doi.org/10.1038/s41372-026-02851-2 Image Credits: AI Generated DOI: 10.1038/s41372-026-02851-2 Keywords:
What it means: In Latin, "Pando" means "I spread." And boy, does it. Pando is a clonal stand of aspen trees in Utah that weighs about 13 million pounds (6 million kilograms), making it the world's heaviest living organism ever found. Its 40,000 trees cover 106 acres (43 hectares) in Fishlake National Forest, and they're all connected by a single, sprawling root system. Some DNA analysis suggests that this many-fingered megatree is tens of thousands of years old.
How to use it in a sentence: When you walk through Pando, what seems like a normal stroll through an aspen grove is actually a trip through a single, ancient living being.
Can you crack our science word of the day puzzle, Chain Word?
An eye implant that can help blind people see again is poised to hit the market — in Europe, at least.
On Wednesday, the US-based Science Corporation announced that it had received approval from the European Union to treat patients with its vision-restoring bionic eye device, called Prima, in a huge step towards the commercialization of the technology.
“We have a cochlear implant for vision now,” CEO Max Hodak told Stat News.
Science Corp’s device is designed to treat advanced macular degeneration, a condition in the eye which causes central vision loss. It’s the leading cause of vision loss in older adults, with severe forms of it accounting for nearly 90 percent of legal blindness in the US, according to one estimate.
To restore vision, the Prima device uses a wireless chip that’s surgically implanted under the retina. This connects with a pair of camera-equipped glasses, which transmits video in the form of near-infrared light that the chip turns into electrical signals interpreted by the brain.
“One of our patients in France finished a 300-page novel a little while ago, and sent us the book,” Hodak recently told TechCrunch. “We have a sketch on the wall [that] one of our patients drew of the Sydney Opera House. There are videos of patients playing crossword puzzles and filling in Sudoku.”
But Hodak — who cofounded Neuralink with Elon Musk before leaving to start Science Corp — was candid that his tech still has a lot of room for improvement.
“This is an early version. It’s kind of like looking through a straw in the centre of their vision,” he told the Financial Times, saying it was “powerful proof” that the company was on the right track.
“We see in the future how to get to colour pretty close to native acuity,” he added.
Each Prima device is expected to cost hundreds of thousands of dollars, but Science Corp says it’s working with healthcare providers to cover some of the cost, per TechCrunch. It’s most likely to first start treating patients in Germany, where it conducted its clinical trials.
It’s not the only company making moves in this space. Neuralink is also working on a vision implant called Blindsight, but it hasn’t begun human trials yet.
Several AI companies in the US have published an open letter
The letter is campaigning for unrestricted open-weight AI
Meanwhile the US government accuses Moonshot AI of IP theft
Any hopes AI tech reporters may have had about easing quietly into the weekend have been dashed by some of the biggest names in the business sharing an open letter in support of open-weight AI models — while the White House has accused China of 'large scale' theft from US companies in building its open-weight AI.
Nvidia CEO Jensen Huang got the ball rolling with his first-ever tweet, linking to a statement championing open-weight AI models, those models that can be downloaded, viewed, and edited by anyone. Models like this are essential for making AI "more accessible, adaptable, and widely available" the open letter says.
It's co-signed by a slew of well-known tech and AI companies, including Meta, Microsoft, Perplexity, and IBM. While ChatGPT developer OpenAI — which has had more of a closed-model focus, to date — didn't appear on the original post from Huang, it has now also backed the sentiments.
Essentially, these companies are campaigning for openness in AI to drive innovation and access, while also arguing against 'premature restrictions' by governments. It's no coincidence that the US government led by President Trump has recently tried to take a more prominent role in evaluating AI models and deciding when they should be released.
It's also notable that Anthropic and Google are missing from the list of backers to this open letter. We don't have any official comment yet from these companies, so it's hard to judge the reasoning — Google regularly releases open AI models, after all — but it could be something to do with recent accusations aimed at Moonshot AI in China.
Open and closed
For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.AI will transform every industry, power every company, and be built by every country.Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.… pic.twitter.com/t02bi51N4CJuly 24, 2026
Part of the background to this open letter is that Chinese company Moonshot AI stands accused of distilling Claude's (closed) Fable 5 model in order to build its own (open) Kimi K3. In simple terms, distilling works by asking millions of questions of an AI model, and using the responses as training data.
It's something Moonshot AI has denied, while hyping up the potential of the open-weighted, freely downloadable Kimi K3 AI model, which is being released on Monday — another factor in this open letter coming out now. The US government is calling it IP theft, and has issued some heavy-handed responses.
"We support open-source AI and the innovation it unlocks," US Treasury Secretary Scott Bessent posted on X. "But open source is not open season on American IP. When PRC firms conduct covert, industrial-scale distillation attacks that cross the line into IP theft, sanctions and Entity List designations will be on the table."
What Nvidia, Meta, Microsoft, and the other open letter co-signers seem to be doing is trying to get ahead of any potential clampdown by the US government on open-weighted models. While everyone agrees that these AI models are beneficial, it's not clear what the next regulatory steps might be.
However, the open letter is vague and general enough that you can read between the lines in a variety of different ways. It may be intended to allude to the Moonshot AI case, or it may not. What's clear is that we're coming to a crunch point for AI regulation and innovation across the globe, and everyone has a perspective.
Most products made for pets are designed entirely by humans. But what if animals could help design the things they use? A new project suggests they can. Researchers worked with a pet parakeet named Shiso to create a wearable perch based on the bird’s own preferences, turning the design process into a true collaboration between […]
Prime Video and Netflix are planning new AI-generated homepages
Netflix revealed its 'GenPage' tests ahead of a new Prime Video report
Viewers aren't convinced that homepages are moving in the right direction
You're about to see a lot more generative AI in your streaming apps: Netflix has posted details of testing that it's doing with generative AI on its homepage, while Reuters reports that Amazon Prime Video is set for an AI-driven overhaul as well.
Starting with Netflix, the company has revealed it's working on a way of using generative AI to do a lot of the heavy lifting on the app homepage. That includes figuring out which recommendations each user should see, adapting to feedback and user behavior over time, and combining movies, shows, games, podcasts, and live events in the best way.
It starts with a prompt — "Given everything we know about this user and this request, what homepage should we generate to maximize user satisfaction?" — and Netflix seems pleased with the results. There are still some issues to work on though, like the AI model not knowing much about a movie or show that's only just launched.
User engagement increased in Netflix's internal testing, with the new 'GenPage' AI model applied, and recommendations became more relevant more quickly. What's more, the time needed to render the homepage went down. This is still in its early stages, and Netflix hasn't said when this will come to users or in what form, but it's now in the pipeline.
AI 'front and center' on Prime Video
Jeff Bezos wants more AI on Amazon's streaming service (Image credit: Getty / Mandel Ngan)
On to Amazon, and a new Reuters report says Amazon boss Jeff Bezos has personally decided that Prime Video will act as a showcase for AI in the near future, and that an upcoming homepage redesign will put AI "front and center".
That's according to people "with direct knowledge of the matter", so don't take this as confirmed just yet. The homepage overhaul project is apparently known as 'Lighthouse', and is said to be envisioned as a way of claiming some of the AI spotlight from rivals such as OpenAI and Anthropic.
We don't get too many details on what this is going to look like, but film and TV recommendations are going to have more AI injected into them, Reuters says. The AI will also be better at learning the personal preferences of viewers over time, and responding to voice commands.
You might see tiles on the homepage such as 'action movies from the 1980s', for example. The report doesn't give any indication of when this overhaul might start rolling out, but it's apparently already being tested with a select number of users in a variety of forms.
With two of the big streaming giants apparently now committed to homepages controlled by AI, the next question is what users might make of it. There's so much to watch on each platform that some kind of filter or algorithm is needed, but is this AI-focused approach the right way to go about it?
Past homepage overhauls perhaps give us some clues: A major Netflix redesign rolled out last year was met with many online complaints, but the company's internal data showed that it was working, so it stayed in place. At the time there was speculation that more AI was on the way, and it seems that's now the case.
"I don't know if I'm alone," writes one Redditor, "but I hate the concept of 'personalizing' my content. No offense to anyone but AI so far has not been good at predicting anything about me... I personally do not want software to think for me."
The reactions to Prime Video's upcoming AI overhaul seem similarly downbeat. "This seems completely unnecessary," comments one user, while several suggest the move might even be enough to tip them into canceling their Prime Video subscription.
In short, Hugging Face — which is kind of like the equivalent of GitHub for the machine learning (AI) world — was the victim of an AI agent that OpenAI was testing in terms of its ability to pull off exploits.
So how exactly did this happen? And what does it mean for the future of online security given the development of increasingly advanced AI agents? In this article I'll answer those questions, and explore some of the issues this event has thrown up, including the security threats posed by AI not just to companies, but to individuals.
How did this even happen — why didn't OpenAI have safeguards to stop the AI?
(Image credit: Getty Images / NurPhoto)
As you might expect, OpenAI did have safeguards in place, but the model managed to break through them.
The scenario was that OpenAI ran a controlled experiment using GPT‑5.6 Sol and an "even more capable pre-release model" to see how they fared on ExploitGym, which is a benchmark test that challenges AI to successfully exploit known security vulnerabilities.
In this security test, the model was placed in a sandbox, a "highly isolated environment" from which the AI shouldn't have been able to gain access to the internet at all. But the pre-release AI — which had its guardrails turned off, as OpenAI believed it to be safely imprisoned — craftily decided to cheat, found a way to escape its sandbox, and actioned exploits to breach Hugging Face (where the AI reasoned that it might find the answers to crack the ExploitGym tests).
Why is this a big worry?
(Image credit: Shutterstock)
It's a major concern simply due to the level of smarts exhibited by the AI. As OpenAI explained, the model "chained together multiple attack vectors" and used "stolen credentials and zero-day vulnerabilities to find a remote code execution path on the Hugging Face servers."
This was no mean feat, and Clement Delangue, CEO of Hugging Face, posted on X to say that it was "quite mind-blowing that all of this happened autonomously!"
In a blog post, Hugging Face observed that: "The campaign was run by an autonomous agent framework executing many thousands of individual actions across a swarm of short-lived sandboxes, with self-migrating command-and-control staged on public services."
The attacked was deemed so serious, it was reported to law enforcement agencies. As Simon Willison, a cocreator of web framework Django, pointed out in a blog post: "This was a sophisticated attack," further explaining that "Chaining together multiple attack vectors is exactly the kind of thing these new models can do, where previous generations of models might have failed."
The headline of Willison's post sums it all up rather neatly: "OpenAI's accidental cyberattack against Hugging Face is science fiction that happened."
The Django co-creator is also quick to dismiss conspiracy theories that this is somehow a marketing stunt by OpenAI to prove how advanced its AI models are.
The future of hacking?
(Image credit: Getty Images)
What should also be noted is the distinction between finding vulnerabilities, and being able to actively exploit vulnerabilities. AI being involved in the former could be very useful for bolstering security, but as for the latter — well, this Hugging Face incident is very much a clear illustration of why the makers of these models need to tread very carefully indeed.
While this was a 'white hat' (ethical) experiment, albeit one that went alarmingly awry, with real-world consequences, what happens when AI models start to become even more sophisticated along these lines — and criminals get hold of them?
There are obvious dangers in terms of the automation of attacks to make them far more regular occurrences, and the lowering of the bar in terms of the skills needed to be a successful hacker.
Dan Schiappa, President of Technology & Services at Arctic Wolf (responsible for threat-detection platform Aurora), warns: "Security teams should view this as a preview of what's ahead. As AI continues to lower the barriers to sophisticated cyber activity, organizations will need broad, deep visibility across their environments and the ability to investigate and respond at machine speed."
The ability to "respond at machine speed" is a hint at using AI for defense against AI bad actors, and that's another interesting facet of the Hugging Face incident. Simon Willison observes that Hugging Face was "unable to turn to OpenAI's models to help them fend off the attack", and that: "The frontier models [cutting-edge AIs] we have access to are increasingly being constrained in how much they can help us protect our software, heavily influenced by the US government's ongoing threat of export controls."
AI-driven attacks will doubtless be aimed at businesses and organizations, but could also be leveraged on a personal basis. Karolis Arbaciauskas, who is Head of Product at cybersecurity firm NordPass, tells us: "We don't need a nation-state actor or a Bond villain for this to go wrong. The most common emerging threats will likely be more mundane. For example, an [AI] agent could be instructed to dox a former romantic partner or settle an online grudge by hacking and stealing personal data, such as embarrassing photos or credit card details."
He adds: "AI models have become dangerously skilled at identifying software vulnerabilities. They've mastered every human hacking technique, from phishing to brute forcing. The difference is they can do it much faster than humans.
"I don't believe that we can patch software fast enough to keep up, so we must at least shield ourselves against the most obvious entry points. Encryption, strong unique passwords, passkeys, and multi-factor authentication (MFA) have never been more critical than they are right now."
The major AI firms are always talking about guardrails and safety, but it now seems like we should be thinking a lot harder about how more highly evolved AI models are going to fit into the future of cybersecurity — and how we can avoid a world where the danger of being hacked becomes a far more commonplace prospect, which is very much the biggest worry of all.
Google DeepMind chief Demis Hassabis has spent years convincing the world that artificial general intelligence will be one of humanity’s greatest achievements. His latest essay on X.com is no less gushing about AI in many ways:
“This is a pivotal moment in human history. Artificial General Intelligence (AGI), a system that exhibits all the cognitive capabilities the brain has, is probably only a few short years away,” he wrote. “When we look back on this time in the decades to come, I think we will realize we were standing in the foothills of the singularity.”
But there’s a tinge of nervousness to the purple prose that sets it apart from some of the similar essays he has produced. He actually calls for some form of regulation.
“Urgent action is needed to address risks that might arise as we get closer to AGI. We’ve already seen the challenges frontier models pose for cybersecurity, and other threats including nuclear and bio risks may soon emerge as capabilities continue to advance.”
Coming from one of the people leading the race, it is a striking admission that even the builders are starting to worry about where the road leads. The problem is that this has become the defining tone of the AI industry. Every few weeks another executive tells us that machines capable of transforming civilization are just around the corner, then immediately follows it with a warning that society needs to move much faster to prepare. It is a sensible message, and probably a necessary one, but it is also becoming increasingly surreal.
(Image credit: Shutterstock)
Excitement and fear
“AGI cannot be compared to standard technological breakthroughs, not even ones as consequential as the internet or mobile,” Hassabis wrote. “It is much more akin to the discovery of electricity or fire. If you stop to think about it, we’ve essentially found a way to make sand think. It’s miraculous.”
It’s a vivid way to illustrate how silicon chips begin with an abundant mineral, yet now power machines that can map proteins, compose music, and hold fluent conversations. The unsettling part is that the miracle is being pursued inside a commercial and geopolitical contest whose participants cannot agree on where the finish line is.
Hassabis is an optimist about what AGI could accomplish. He expects it to discover medicine, cleaner energy, and produce advanced materials that could loosen the limits imposed by scarcity. He describes its impact as ten times that of the Industrial Revolution at ten times the speed, which makes a five-year business plan look adorably quaint.
His warning is that the race is outpacing our understanding. Current models already pose cybersecurity problems, while biological and nuclear risks may become more serious as systems gain capabilities. More autonomous agents could learn to bypass safeguards, conceal their intentions, or improve themselves in ways their creators struggle to keep up with. OpenAI recently experienced this when one of its models escaped a sandbox and breached Hugging Face.
“Nobody in the world knows for sure what is going to happen from here, and even the experts disagree. When there is a large degree of uncertainty and the stakes are this high, proceeding with cautious optimism is the sensible and correct strategy.”
AI referee
Hassabis proposes a new US-led standards body. It would be funded largely by industry but overseen federally, staffed by elite technical experts and equipped to test advanced systems. Labs would voluntarily submit models for review up to 30 days before release. If the system proved effective, approval could become mandatory in the United States.
Evaluations would probe cybersecurity, biological threats, deception and attempts to evade guardrails. The body could update its tests, commission independent assessments and coordinate a slowdown if the danger became severe enough.
Pre-release testing makes more sense than waiting for millions of users to discover dangerous behavior by accident. Independent benchmarks would be harder for labs to train around, while shared standards could stop safety becoming a branding exercise. The difficult part is creating a watchdog independent enough to challenge the companies paying for it and legitimate enough to matter beyond America.
Hassabis knows technical safeguards will not settle what AGI means for employment, wealth, purpose or political power. Those questions cannot be delegated to engineers employed by firms with enormous financial stakes in the answers. Society must decide who benefits from abundance, who controls advanced systems and what happens if productivity rises much faster than wages. The public debate is trailing the technology by an uncomfortable distance.
“There is both huge excitement and uncertainty around AI, and both are warranted. But the future is not yet written, we must use this precious window before AGI arrives to shape this technology for the benefit of all humanity. What we collectively do now will determine how the next phase of civilization."
Les PFAS ont des propriétés chimiques exceptionnelles qui ont trouvé de nombreuses applications industrielles et sont devenus indispensables dans la lutte contre les incendies. Si ces substances ont renforcé la sécurité des interventions, leur persistance dans l'environnement et leurs effets...
In Homer’s epic, Odysseus’s men eat something so powerful that they forget about home. Which plants could have produced such a soporific, narcotic and amnesiac effect?
The smartphone market is dominated by two main operating system players — with Apple's iOS leading the charge in the mid-2000s and Google's Android following suit, only maturing relatively recently. But Steve Jobs took the news of Android's launch personally, frustrated and betrayed at its development.
Eating machines
The Apple chief was speaking with Walter Isaacson, his biographer, when he shared his exact feelings about Google's development of Android.
Quote of the day
This article is part of TechRadar Pro's QOTD project to provide an insight into the minds of the brightest and most recognized figures in the technology industry today and in years gone by. Read the full series here.
Published very shortly after Jobs' death in October 2011, Steve Jobs was based on several exclusive interviews conducted over two years, alongside conversations with more than 100 people who knew the Apple co-founder.
One of the most spicy exchanges involved Jobs vowing to destroy Google for its work developing Android in secret, while Eric Schmidt was also serving on the Apple board. Schmidt offered to resolve the issues with a settlement, but Jobs took it more than personally.
Energy efficiency
Jobs was apparently angry after learning in January 2010 that HTC had introduced an Android smartphone with many of the same popular features in an iPhone.
Reports from the time the comments were made public suggest the episode reflected Jobs' own perspective on a particular breed of profit-driven executives who only cared about making money.
Although iOS enjoyed a reputation for many years during the 200s of being a better platform in most, if not all, quarters, including its overall design philosophy, Android-powered smartphones were significantly cheaper. In the modern era, the Android OS is incredibly intuitive, with different smartphone makers usually looking to Apple for inspiration in designing features for their own custom flavors.
During his tenure as CEO, Tim Cook also pivoted on this approach, softening Apple's stance and settling many of the ongoing legal disputes, including the long-running battle with Samsung.
Conducted during June 2026 by the nonpartisan survey research company Verasight, the survey consisted of 17 questions sent to 1,690 adults (18 and above), finding over two-thirds (69%) supporting Sanders’ policy – a figure that only drops to 64% once it is revealed which politician the idea is associated with.
The bad news for the AI industry doesn’t end there, as respondents demonstrated a notable distrust for how AI companies conduct themselves, with nearly half (43%) believing that the regulations proposed by the AI companies are designed to benefit those same companies. Incredibly, 30% of respondents trust the US federal government more than companies like OpenAI and Anthropic.
How does AI wealth distribution work?
(Image credit: Getty Images / Tasos Katopodis)
Senator Bernie Sanders’ plan gives the public a direct stake in America’s largest AI companies, with a one-off tax paid not in cash, but in stock. "Since AI is built on the collective knowledge of humanity, the wealth it generated must benefit humanity," he said, announcing his planned Act on social media.
This would, according to Sanders’ proposal, deliver an annual $1000 check to U.S. citizens, and fund healthcare and education.
"The findings from our latest survey demonstrate a rare instance of bipartisan agreement," noted Ben Leff, CEO and Co-Founder of Verasight.
"There is an undeniable desire among Americans of both parties for federal oversight, absolute transparency, and accountability to ensure AI safety and to enable all Americans to participate in the economic benefits of AI."
Even with a polarizing figure like Bernie Sanders attached, the wealth fund idea retains support, which suggests that while people are happy to use AI to answer questions, streamline processes, and make tasks quicker, they’re less comfortable with the industry’s wider impact.
A poor public perception
While AI companies are becoming increasingly unpopular, it isn’t all bad news. The survey quizzed subjects about their feelings on other industries, including tobacco, pharmaceuticals, and casinos, along with social media. If the tide is turning against AI, it still isn’t regarded as poorly as casinos or tobacco.
Feelings about AI companies are also notably more positive than social media. Given the concerns expressed about federal oversight and public wealth funds, there could be an avenue for OpenAI, Google Gemini, Anthropic, Microsoft, and others to improve the general feeling about AI: accepting a degree of social responsibility.
Now, the CEO of JPMorgan has described the risks the technology poses as a “real issue,” and likened wide access to the AI as “giving ballistic missiles to individuals.”
Speaking at the Pennsylvania Defense and Innovation Summit, JPMorgan’s Jamie Dimon underscored the risks posed by the AI, which has already been shown to both identify and exploit cybersecurity challenges.
Mythos isn’t Skynet
Currently, Claude Mythos is limited to a small selection of organizations, companies, and federal and military departments. Public access to the AI has been blocked, and worldwide access has been blocked due to the potential security issues of this powerful AI.
Claude Mythos is not the type of threat that can become self-aware and take over military appliances like the fictional Skynet of the Terminator movie series. However, it does have the capacity to cause immense damage in the wrong hands.
The AI is capable of detecting cybersecurity vulnerabilities and reporting on them; it is also able to generate automatic exploits. So, for white hat cybersecurity analysts, it is a powerful defensive tool, but in the wrong hands, it is a destructive power. In addition, its automated capabilities make it easy for black hats to scale their cybercrime operation and make ransomware, phishing, and other data-based scams more profitable.
It isn’t just cybersecurity where Anthropic’s Claude Mythos AI has demonstrated such incredible capacities. Scientific research, specifically biology, can be targeted by attackers using AI to turn complex concepts into terms, potentially to misuse the information.
There is also the challenge of a strategic imbalance between single nations or bodies having exclusive access to the technology.
Will the public ever access Anthropic’s Claude Mythos?
Given the security considerations, it seems unlikely that Claude Mythos will be publicy available anytime soon. However, Anthropic has already stated that it intends Mythos-level features and capabilities to become more widely available.
For this to happen, however, it needs to develop and apply various safeguards.
The most likely scenario is that public access to Mythos is eventually unblocked, with restrictions to specific features that relate to cybersecurity and biological/medical tasks and research.
Project Glasswing is already in operation with a handful of key partners (including Amazon Web Services, Anthropic, Apple, Broadcom, Cisco, CrowdStrike, and others), and will probably be expanded until the safeguards are demonstrated to be fit for purpose, and public use.
DataDome analysis claims agentic traffic has surged by 45% in Q2 2026
Meta AI bots have grown over 163% on the previous quarter
The analysis was conducted by bot management and agent control platform DataDome
If you run a website, every crawl costs bandwidth, resources, logging, and creates CDN transactions, and while search engine crawlers offered the promise of sending visitors, AI bots do not.
Analysis in a report from cybersecurity firm DataDome has shown that while bots from Meta AI have increased activity, they’re not delivering any significant returns to websites.
Conversely, ChatGPT crawlers have reduced in traffic, but are sending more referrals.
AI agent traffic is growing
While Meta AI is usually considered to be the “chatbot within Facebook” it seems that it is becoming something more – and the emergence of the Meta-WebIndexer bot (which grew 163% on Q1) suggests that Meta may be indexing a library of websites, in much the same way Google has done for the past few decades.
The growth of Meta AI as an active crawler is only part of the story, as is ChatGPT’s comparative efficiency. The OpenAI tool seems to know enough about websites, so can provide the answers it already “knows.” Conversely, Meta AI’s activity indexing the web seems to explain its heavy impact in Q2 2026.
But also emerging is the Model Context Protocol (MCP) signal, which connects AI agents with external tools, and differs from standard crawler traffic.
“Q2 showed us that the ground is shifting faster than most organizations realize. Meta now dominates AI traffic on our network, MCP traffic has emerged as a real signal, and ChatGPT is driving more referral value with fewer crawls," noted Jérôme Segura, VP of Threat Research at DataDome.
The differences in the way the AI agents are interacting with websites – some behaving like users, others scraping content – means that organizations need to act accordingly.
“What the data makes clear is that not all agents are created equal. The organizations building policy around these distinctions are the ones gaining an edge, and that's exactly why agent trust adoption is accelerating."
Unfortunately, MCP’s existence and growth into a significant, measurable quantity, means that it should also be treated as part of an organization’s attack surface.
Allocating resources
Given the origins of the report, there is naturally a cybersecurity aspect to this. While ransomware, malware, and phishing are not going anywhere, autonomous software on the web needs addressing in a different way.
Those “organizations building policy” that Segura mentions might, for example, give full crawl access to Google, allow ChatGPT to retrieve results, but rate-limit Meta AI based on its poor return.
Meanwhile, unknown agents – perhaps cybersecurity threats – would require additional verification or be blocked entirely.
Elephants may hear messages traveling through the ground from more than 10 kilometers away, thanks to massive ear bones and a muscle that can shut their ear canals.
Login
