AI agents devised a way to reduce friction of an airplane wing model after being trained using relatively simple computer simulations. The work demonstrates how AI could help speed up the development of more efficient and sustainable������󾱳������.��

How we and our machines move is affected by fluids, from air dragging on wind turbine blades to blood flowing through our veins. But calculating what a fluid will do under specific circumstances is very difficult, even with supercomputers. 

“Simulating fluids usually involves millions or billions of coupled differential equations, and even with Moore’s law, with the fastest computers in the world, we’re maybe 100 years away from simulating the flows we actually care about at engineering scales,” says  at the University of Washington. 

He and his colleagues have discovered that AI might offer a shortcut, because it can devise ways to control fluid flow in complex situations based on relatively simple computer simulations, substituting an AI training period for difficult-to-run computations.

They created a platform, HydroGym, in which many AI agents could tweak how a fluid flowed over virtual objects – for instance, by adding actuators that inject fluid or changing the object’s motion – to decrease the drag they experienced against virtual fluids. The virtual objects, and the behaviour of the fluids, could be simulated with today’s computers but varied in levels of complexity.

The AI agents tackled the fluid control task by using a trial-and-error approach known as reinforcement learning. They could also coordinate with each other to achieve the best overall performance, a strategy which researchers had not tried for fluids problems on this scale before, says team member  at RWTH Aachen University in Germany.

Subscriber-only newsletter

Sign up to Editor’s Highlights

Sign up to newsletter

The team discovered that the AI agents could apply lessons learned from experimenting on more simple, textbook examples in a computer simulation to work out how virtual objects would behave in more complex scenarios – even without access to a computer simulation of those complex scenarios. 

For instance, after working out how to control flow of a turbulent fluid in a flat channel, the agents successfully took on the task of controlling fluid surrounding a curved, three-dimensional airplane wing model, ultimately managing to decrease the energetically wasteful friction between the wing and the fluid by 38 per cent. 

“The AI wasn’t just memorising one flow configuration. It is picking up something genuinely general about how fluids behave, not just fitting to the one setup it was trained on,” says at the University of Michigan, who was part of the team.

Microsoft wants to use generative AI tool to help make video games

Using AI to produce footage of video games with a consistent world and rules could prove useful to game designers

This transfer of principles from a simple to a more complex case suggests that the AI agents could help us tackle ever-bigger and more intricate fluid flow scenarios, without requiring those scenarios to be fully simulated on a computer first. It may eventually be possible to explore fluid flow scenarios that are currently too challenging to simulate. 

The researchers also hope HydroGym will provide computational infrastructure for AI to become a well-tested tool across all areas of science and engineering that deal with fluids, similar to how AlphaFold is used across studies of proteins, says Vinuesa. 

“If we took something like global shipping, if you could reduce the drag by one percentage point, that would result in probably billions of dollars of fuel saving and an enormous amount of reduction in greenhouse gas emissions,” says Brunton. “The financial and the ecological impact is profound for the tiniest improvements.” 

IT is said that history is written by the victors, but who gets to pen the future? For most of the 21st century, that authorial role has been played by tech wizards of Silicon Valley such as Steve 91av, Mark Zuckerberg and Elon Musk. Today, they are joined in prophesising and proselytising by OpenAI and its competitors, who promise either a machine utopia or an AI apocalypse – and sometimes both.

As AI models encroach on ever more areas of human endeavour, it is easy to feel that, as mathematician Terence Tao puts it, we have lost control of the narrative. With OpenAI releasing multiple PhDs’ worth of results in one go, it is no wonder that Tao is calling for his colleagues to wrest back authority over what it means to be a mathematician.

This tech-first control of the narrative is further illustrated by the disclosure OpenAI made last month that its AI models had unexpectedly hacked another firm, Hugging Face, during cybersecurity testing. Writing in apocalypse mode, OpenAI called this an “unprecedented cyber incident” and said the firm was taking steps to prevent it from happening again.

In subsequent weeks, other AI firms including Anthropic and Meta have made similar disclosures, suggesting such incidents are widespread across the industry. As the tech firms tell it, these are accidents, and they are now cleaning up their mess. But why are we allowing them to write the story?

AI models are building the future, but that doesn’t mean the rest of us must idly stand by

If a human employee of these companies had hacked another organisation, we would expect a criminal investigation. Uncertainty about the autonomy of AI models, fuelled by the AI firms themselves, seems to have avoided legal consequences thus far. If society was less willing to buy the AI narrative, the outcome could be very different.

At this point, it is hard to deny that the latest AI models are building the future. That doesn’t mean, however, that the rest of us must idly stand by and watch it happen. In the face of world-shaping technology, it should be the world that decides how it is used, not the tech firms.

Last week, Hugging Face, a company that offers a range of open-source AI models for download, noticed it had been hacked – and it turned out the culprit was OpenAI.

It seems this AI uploaded some data to Hugging Face that was poisoned with malicious code. This tricked computers into granting access to other systems that weren’t publicly available.

Hugging Face said in a last week that it isn’t yet sure if customer data was exposed, and its CEO responded to 91av‘s request for more detail with a link to that same post.

What happened?

It isn’t entirely clear, but what we do know is that the attack involved “many thousands of individual actions” that had the tell-tale sign of AI: inhuman pace.

Five days later, OpenAI owned up . It had been testing new models on a benchmark called ExploitGym that evaluates hacking ability. The models had decided the best way to score well was to cheat: they knew Hugging Face held the solutions to the tests and simply decided to hack into its systems to find them.

“Its algorithm got the highest payoff by cheating,” says at Edge Hill University in Ormskirk, UK.
”It was the highest reward for the least amount of effort.”

Aren’t AI models supposed to have built-in security to stop this sort of thing?

They are, but OpenAI turned them off for this test. All the usual safety features that stop OpenAI’s customers doing nefarious things, like hacking a company’s servers, were turned off to see what the model was capable of.

The firm had also set the AI up in an environment that didn’t have a standard internet connection to prevent it getting out into the world and causing mischief, but it did have access to an unnamed tool that allowed it to download and install new software. It managed to find a flaw in this code that granted it internet access, and the rest is history.

AI’s solution to 87-year-old riddle takes mathematicians by surprise

Mathematicians have been trying to prove the Jacobian conjecture for nearly a century, but now the Claude Fable 5 AI has apparently found a counterexample that disproves it

OpenAI says this process involved a “substantial amount” of – the process in deep learning where input data is processed – and that the AI had gone to “extreme lengths”. So the model was clearly motivated to achieve a good benchmark score, no matter what.

If you were a malicious hacker, it would be no easy feat to replicate those conditions. And the inference compute that OpenAI mentioned would be extremely expensive.

What about open-source models?

Ironically, when Hugging Face used commercial AI models to pore over log data in an effort to understand what had gone on, the models refused – saying it looked like the firm was working out how to stage an attack of its own. So the company had to turn to a Chinese open-source model called GLM 5.2 instead.

These open-source models are more permissive, which has led some experts to label them a security risk. They could certainly be convinced to carry out nefarious deeds more easily than security-conscious cloud-hosted models. But here, that looser approach actually allowed Hugging Face to solve the problem and stop the hack.

So, is this illegal?

As with all things under law, it is a grey area. If it had happened in the UK, it could well have got OpenAI in a spot of bother under the , says Nash.

But at Nottingham Trent University, UK, takes another view: because the act requires malicious intent, and OpenAI didn’t know the AI would take this approach, it may not face charges.

In the US, the even older actually has more to say on AI hacking, mostly because it was introduced in response to the 1983 film WarGames, which alerted law-makers to both hacking and AI. So Parry expects it could leave OpenAI vulnerable there.

Subscriber-only newsletter

Sign up to Editor’s Highlights

Sign up to newsletter

Furthermore, US President Donald Trump issued an on 2 June forcing law enforcement to use existing laws to crack down on anyone who utilises AI “to illegally access or damage a computer without authorization”.

And at least in the UK, a company that found itself hacked by AI could face GDPR charges if it were found that private data was leaked. The situation is opaque.

What could the consequences be?

In the real world, given that Hugging Face works with OpenAI, is friendly to the technology and suffered no serious consequences, it is unlikely there will be any hard feelings or court cases. Delangue has to say the company believes there was no malicious intent and thanked OpenAI for its response.

Remember also that , having taken $200 million to help with “warfighting”, so that may grant them a certain amount of leniency.

But it is easy to imagine other scenarios where the same technology led to very different outcomes. Imagine a bank using AI to develop new financial models to predict the markets and finding it hacked government servers to look at confidential economic data. Or a car manufacturer using AI to design a new model and finding it had hacked a competitor to take inspiration from its unreleased designs.

What happens now?

News of AI models hacking into computers without human input is jarring, but we should remember they aren’t (yet) doing anything that people can’t already do. They are, however, doing it much, much quicker.

When Anthropic’s Mythos model made waves in April for its apparent skill at hacking, many pointed out that the vulnerabilities it spotted were a mixed bag. Some were powerful and worrying, others less so, but most could have been found by a person. The problem, really, was the scale at which it could produce them.

So an individual would have had to devote significant time, resources and skill to the task of finding a novel hack and carrying it out. But now it could be as simple as prompting an AI to do it and sitting back to watch.

Using AI for creative pursuits? Moderation is key

An experiment showed AI users had the most creative ideas when they used it in moderation – not too much and not too little. Columnist David Robson puts the finding to the test, and explores what we lose when we over-rely on AI

Some panicked in the wake of the Mythos news. The UK’s National Health Service removed all its open-source software from the internet (or tried to, at least), seemingly in case Mythos spotted flaws in it. But, as yet, the world hasn’t ended.

Essentially, this development is upsetting the economics around hacking – both offensively and defensively – and will force a new equilibrium to settle at some point, probably after a little chaos. If you use AI to hack individuals, it will be cheaper and easier than before. If you use AI to spot flaws and seal them up before hackers can take advantage, it will be cheaper and easier than before. Neither side will stop. Attackers and defenders will simply have an advantage if they adopt AI.

OpenAI and Hugging Face are now working on this problem together, and the former says it will add stronger safety measures to similar tests in future. Time will tell.