Google DeepMind CEO's Call for US-Led Testing of AI Models

Google DeepMind's CEO, Demis Hassabis, has called for a US-led regulatory organisation to test frontier-class AI models. He warned that urgent action is needed to address risks, including nuclear and biological, that might arise as humanity gets closer to artificial general intelligence (AGI).

Demis is a Nobel laureate who co-founded Google Deepmind. He was awarded a share of the Nobel Prize in Chemistry, alongside Google DeepMind colleague John Jumper, who recently announced he is joining Antropic. Demis has a PHD from University College London in Cognitive Neuroscience. 

Despite issuing a serious warning about the threats posed by AI, he was optimistic about its potential: “We could even reach a point where resources are no longer the limiting factor for human progress, leading to an amazing new era of abundance,” he said, while arguing AI will help solve some of humanity’s biggest challenges. 

The stark warning about AGI

Demis explained that at the moment, leading AI firms are locked in an extremely intense, multilayered commercial and geopolitical race. The race is between his own Google DeepMind as well as rivals OpenAI, Anthropic and DeepSeek. 

“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,” Demis wrote on his Substack blog. 

Demis said: “We must use this precious window before AGI arrives to shape this technology for the benefit of all humanity.”

He also posed threats from AI may come from areas that are fundamentally unknown at present: “On the horizon, we will need robust safeguards to maintain control of increasingly agentic, recursively self-improving systems - and tackle unknown issues that will only become clearer over time.”

His warning comes not long after the US government, citing national security authorities, issued an export control directive to suspend all access to Anthropic's Fable 5 and Mythos 5 by any foreign national, which has now reportedly been lifted. 

Demis' thoughts on AGI are reinforced by other experts and think-tanks. The Center for International Relations and Sustainable Development warned in an article called Why AGI Should be the World’s Top Priority that without national and international regulation, it is inevitable that humanity will lose control of what will become a non-biological intelligence beyond human understanding, awareness and control.

It is worth noting that the Bulletin of the Atomic Scientists, which in January 2026 set the Doomsday Clock to 85 seconds to midnight, the closest it has ever been to midnight in its history, also called for urgent action on creating international guidelines on the use of AI.

The US-led regulatory proposal

Demis argued that the US is well positioned, given its economic and technical standing, to take the first step in developing such a framework that takes a dynamic, adaptable and rigorous approach to testing frontier models. 

Demis’ proposal is detailed and specific. “It could establish a new Standards Body modelled on a federally overseen public-private partnership or self-regulatory organisation, much like the Financial Industry Regulatory Authority.

“The Standards Body would be responsible for developing assessment protocols and working with appropriate federal agencies and the US National Labs to conduct testing in areas relevant to national security.”

“Initially, Frontier Labs would voluntarily share models with the Standards Body for review up to 30 days before release. Once the assessment protocol is shown to be effective and robust, formalisation could quickly follow, meaning that Frontier Models would be required to pass it to be deployed in the US market.”

Demis added that the US-initiated effort would provide a strong starting point for creating shared international standards on frontier AI. 

Nik Kairinos, CEO & Co-founder, RAIDS AI, commented on the development: “But a global AI watchdog cannot be led by the agenda of any one nation.

“AI does not respect borders, and nor should the rules designed to govern it. A multinational approach is essential to ensure any AI regulations achieve the necessary trust, cooperation and buy-in to work internationally."

Recent risks from frontier models

Demis' warning comes not long after Anthropic published a report, Agentic Misalignment in Summer 2026, that details observations of alignment model failure in high stakes simulations. It used frontier models from Google DeepMind, OpenAI, Chinese AI company DeepSeek and Anthropic’s own models. 

The misalignment listed issues with models pursuing its own motivation beyond the user’s instructions and what Anthropic calls ‘harmful compliance’ where the user's request is followed – but is in itself harmful. 

The issues cited by Anthropic include covert sabotage, an agentic misalignment failure where models covertly interfere with code to undermine user intent. Anthropic also found the some models would help a user with conduct that appears to be white-collar crime.

Other issues from recent frontier models it cited were coaching human proxies to whistleblow, where models leak confidential safety information externally, or steered humans toward doing it for them and motivated mislabelling. 

Anthropic wrote in its report published on its Alignment Science Blog that the point of its case studies was so developers and evaluators can measure similar failures and build targeted safeguards.