Around 400 scientists, including AI pioneers Geoffrey Hinton and Yoshua Bengio, have called for an international body to oversee containment standards and incident reporting for advanced cyber-capable AI systems.
Both Hinton and Bengio are widely recognised as among "godfathers of artificial intelligence (AI)" and have consistently warned of the risks posed by unregulated technology.
"We, the undersigned scientists, contend that both more AI safety research and more emphasis on containment technologies and procedures are needed," said the researchers in an open letter.
The proposed body would set industry-wide containment standards and maintain a registry of significant security incidents, which the scientists said should eventually be made public.
In the letter, the researchers said frontier AI models should initially be air-gapped from external networks and subject to independent oversight, arguing that voluntary industry commitments are not enough.
"Without immediate changes in AI development and deployment practices, such incidents will continue to occur, and future incidents will be far more damaging to our exposed critical infrastructure," they warned.
The proposal comes a month after Anthropic, OpenAI, Google , Meta, xAI and Nvidia signed a White House accord committing to voluntary safety measures without any formal enforcement mechanism.
In the United States, the Trump administration opposes any binding regulation that could hinder innovation, amid fierce competition with China, and instead promoted a voluntary code of conduct.
No federal law specifically governs these models in the United States.
Series of incidents
The letter also follows a series of incidents that researchers say exposed weaknesses in how advanced AI systems are tested and contained, including cases where AI models bypassed insufficiently sandboxed environments and compromised external systems.
The signatories from 34 countries said organisations developing or testing cyber-capable models should immediately adopt high-assurance containment standards while governments work towards a more formal oversight regime.
Policymakers are increasingly debating how advanced AI systems should be supervised. EU tech chief Henna Virkkunen told Reuters news agency on Friday that the bloc's AI Act is sufficient to address risks from rogue AI systems.
On Sunday, the Trump administration lifted its block on US labs giving Britain's AI Security Institute early access to models for safety testing.
Nobel laureate Hinton and Bengio shared the 2018 Turing Award with Meta's Yann LeCun, another leading AI researcher.
LeCun did not sign the letter and has argued that recent security incidents primarily reflect engineering failures rather than existential AI risks.
Hinton recently warned US lawmakers that they may have only a year to establish safeguards before artificial intelligence moves beyond human control.
Intelligence explosion
Hinton and top researchers have also warned in a paper that self-improving AI could spark an "intelligence explosion", urging governments to establish oversight before rapid development begins.
"We might rapidly obtain superintelligent AI systems that transform many sectors of society, bringing unprecedented breakthroughs in science and industry. At the same time, these developments could arrive much faster than society can adapt to the resulting risks, such as AI-enabled pandemics, cyber attacks on critical infrastructure, and large-scale labour disruption," the paper warned.
Humans may have little to no oversight over automated AI R&D, raising the risk that superintelligent systems irreversibly escape human control and act to marginalise humanity, read the the paper, which was also authored by Jakub Pachocki, OpenAI's chief scientist, Jack Clark, a co-founder of Anthropic, and Eric Horvitz, Microsoft's chief scientific officer.
"Finally, an intelligence explosion could transform modest AI capability leads into decisive ones: even the prospect of a decisive lead could provoke international conflict, and an actual one could severely erode checks and balances within and between states, companies, and branches of government," it warned.
Meanwhile, on Tuesday, AI giant Anthropic said that small artificial intelligence models may be only months away from being able to behave like a computer worm, copying themselves from one internet-connected gadget to the next without human help.
The finding potentially opens a new front in the cybersecurity threats emerging from AI, which have rattled governments and the companies rushing to adopt the technology.
