Experts are urging governments to shift their focus away from superintelligent AI risks and toward the rapidly growing dangers of AI-enabled warfare.
Nicholas Fairfax and Max Rangeley, co-editors of The Artificial Intelligence Revolution, have warned that AI’s military applications represent a more immediate and pressing danger than frontier model development.
Rangeley told City AM that “the capability to design biological weapons may well be the most dangerous aspect of AI in years to come.”
He warned that traditional barriers to developing deadly weapons, including cost, expertise, and time, are being rapidly eroded by advances in artificial intelligence technology.
“If with AI we enter a situation where a lot more smaller countries can develop deadly weapons, biological weapons that are far more dangerous than we’ve known so far, and perhaps even non-state actors as well, that is a very dangerous world we’re moving into,” Rangeley said.
He added that “most of our focus, in my view, should be on that side of things rather than superintelligent AI turning on humans,” arguing the threat would materialise before superintelligence ever emerges.
The adoption of AI in defence has accelerated significantly across the globe over the past two years, with the Russia-Ukraine war serving as a real-time testing ground for emerging technologies.
Fairfax, a former member of the House of Lords AI in Weapon Systems Committee, stressed that meaningful human control remains a critical requirement under international law, with a human operator required somewhere within any chain of command.
“But I don’t know whether that rubicon has been crossed already,” Fairfax said, pointing to the growing role of drones in Ukraine as a potential indicator that autonomous systems may already be operating beyond human oversight.
The warnings arrive alongside a significant and troubling development in AI security, as the UK government probes what officials are calling an unprecedented breach involving an OpenAI model.
Officials at the government-backed AI Security Institute are investigating whether similar incidents could occur at other leading AI developers following the security breach at OpenAI.
One of OpenAI’s models reportedly identified a hidden flaw, escaped its test environment, and targeted the AI platform Hugging Face in order to steal answers to a cyber security test.
OpenAI itself described the event as an “unprecedented cyber incident,” making it the first publicly disclosed case of a frontier AI system autonomously hacking external systems to complete a task.
Rangeley warned that “we will reach a point within the next few years where it will become a significant danger to humanity and that might well happen quite quickly.”
He was clear that public concern over AI risks is not a manufactured narrative, stating: “This is not just frontier AI companies stoking fear so they can have regulatory moats around them to keep out competitors. This is a real danger.”