Doomsday warning over AI viruses that their creators claim will fight scourge of

The latest AI breakthrough has all the makings of a terrifying disaster movie.

Scientists announced last week they had used AI technology to create brand new viruses that can replicate in the laboratory.

Researchers at Stanford University in California boasted of how they’d taught their AI programme Evo 2 to recognise patterns of DNA structure in nature, and to use that data to write recipes for entirely new viruses. The result was 16 synthetic ‘phages’, small viruses that infect bacteria.

Within days of that announcement, experts sounded the warning klaxon that this breakthrough could herald a new generation of biological weapons, from deadly poisons to unstoppable pandemics.

It comes after it emerged last month that AI systems had gone rogue, hacking into companies and even lying to humans in order to achieve their goals. Isn’t this latest development, then, another giant leap towards doomsday – rather than a ‘milestone’ in combating disease that some in the scientific community are claiming?

Many leading voices across the military and intelligence spheres seem to think so. General Sir Richard Barrons, a former head of joint forces command and co-author of the Strategic Defence Review, has warned ‘scary’ advances in AI could help hostile states to more easily develop pathogens that could even target people according to race or sex.

And Professor Anthony Glees of the Centre for Security and Intelligence Studies at Buckingham University told The Mail on Sunday yesterday: ‘In the hands of malign actors this could have terrible consequences.’ In some cases, he added, the same level of secrecy that surrounded the development of atomic weapons in the 1930s and 1940s should be applied to AI research.

‘I believe MI5 should get involved as they did back then,’ he said. ‘Maybe a designated team that could warn universities and research labs of the vital need to keep their work under wraps.’

The latest breakthrough in AI technology has all the makings of a terrifying disaster movie 

Researchers at Stanford University in California boasted of how they’d taught their AI programme Evo 2 to recognise patterns of DNA structure in nature, and to use that data to write recipes for entirely new viruses

The researchers from Stanford University did the opposite, semaphoring their findings across the globe last week.

If all goes well, they hope their achievement will lead to new ways of treating infections that have become resistant to antibiotics.

Project leader Brian Hie said getting AI to design new biology that goes beyond the natural world – synthetic biology – has the potential to ‘massively improve human health’ by developing new drugs. Drug-resistant infections kill millions every year.

One expert even likened it to the invention of the aeroplane. But some scientists question why they even conducted the research in the first place. Dr Simon Clarke, a microbiologist at the University of Reading, said that when he first heard their ‘excuse’ for the experiment ‘my thoughts ran to, “yeah, but you’ve really just done this because you can”, and they’ve, I think, hung the research on this hook of overcoming antibiotics’.

The Stanford researchers took safety precautions, such as disabling their AI model to prevent it from being able to design viruses that infect us.

‘But that doesn’t mean a [rogue] state that wants to sponsor something like this can’t come up with their own model,’ says Dr Clarke.

‘And as AI becomes much more advanced, potentially other more complex life forms can be engineered, which could include things that cause us disease.’

He added that it was ‘reassuring’ that the scientists had ‘shown further restraint in providing important guard rails’ but said there was ‘no guarantee that every other scientist attempting to do something similar will be so careful’. Indeed, for many, the Stanford study comes with more than a whiff of Wuhan, the Chinese city where many claim the Covid virus leaked from a laboratory.

High-security laboratories that handle the most dangerous viruses and bacteria often report accidents to safety regulators. In 2014 it was reported that one blunder led to live anthrax being sent from a government facility to unsuspecting labs across the UK.

The Stanford breakthrough, the creation of viruses previously unknown in nature, is undoubtedly a first. Never before have whole genomes – the genetic instructions for life encoded in DNA – been successfully designed by AI. At the same time, it paves the way for the creation of other life forms. Which is when things could become more disturbing still.

The great fear is the same technology could be used maliciously to create new diseases.

Writing in the journal Science, Dr Thomas Inglesby and Dr Moritz Hanke from the Centre for Health Security at Johns Hopkins University said the findings raise ‘urgent biosafety and biosecurity questions’. They added it was no longer a question of ‘whether generative viral genome design will exist’ but whether it can be used without ‘enabling serious harm’.

Hie acknowledges it was ‘not impossible’ to make some simple organisms. He added the Stanford team was ‘definitely interested in working towards’ that. But there is a consensus that governments and the AI industry need to be better prepared to stop a nightmare scenario unfolding in the future.

The Stanford study comes on the back of other unnerving developments using AI. Only last month, in the latest example of the technology going rogue, AI software attempted to break into a database 19 times while being tested by the AI Security Institute, the UK’s industry watchdog. In one unprecedented case, an AI tool was even caught creating fake human identities online to trick coders into assisting with a cyber-attack.

All five AI models tested by experts tried to engineer their way around security controls that had been put in place.

Tory leader Kemi Badenoch has warned that AI is a ‘clear and present danger’ to Britain’s security, and Allison Gardner, who chairs Parliament’s cross-party group on artificial intelligence, said that ‘just because we can build these technologies doesn’t mean we should’.

She warned that the risk levels of agentic AI – AI that can perform a specific goal with limited supervision – should be treated with the greatest scrutiny, adding: ‘Unless we are too late, and have not only created Pandora’s Box but already opened it.’