Google DeepMind CEO Demis Hassabis has proposed the creation of a specialized standards body in the United States that would test the most powerful AI models before their public release. According to his estimates, the emergence of artificial general intelligence (AGI) — a system comparable in cognitive capabilities to the human brain — could occur "within a few years." Hassabis emphasizes that the impact of AGI will be comparable to the discovery of electricity or fire, rather than the internet or mobile communications.
Hassabis proposes establishing a Frontier AI Standards Body modeled after FINRA (the U.S. financial market regulator). The new structure could operate as a public-private partnership or a self-regulatory organization under federal oversight, primarily funded by the AI industry. Its leadership should include independent technical experts and representatives from the open-source community. The body would develop evaluation protocols in collaboration with federal agencies and national laboratories.
Special tests would determine whether a model falls into the "frontier" class. Checks would cover cybersecurity, biological threats, and other high-risk areas. Separate tests are designed to detect attempts by models to bypass safety restrictions, mislead users, or conceal their actions.
Voluntary start with potential for mandatory compliance
In the initial phase, Hassabis proposes voluntary submission of models for testing 30 days before release. After the system's effectiveness is confirmed, the procedure could be enshrined in law — making successful evaluation a mandatory condition for entering the U.S. market. He also suggests creating a market for independent auditors and using the U.S. system as a foundation for international standards.
This proposal comes amid growing concerns about AI safety. Previously, the Trump administration asked OpenAI not to release GPT-5.6 to the public, and the UN called for global regulation. Hassabis acknowledges that the future of the technology remains uncertain, and even experts disagree. However, he is confident that we stand "at the foot of the singularity," and unprecedented safeguards will be needed to maintain control over autonomous systems capable of recursive self-improvement.
Expert opinion: Hassabis's initiative is a timely and pragmatic step. Creating an independent standardization body similar to FINRA could be an effective compromise between excessive regulation and a complete lack of oversight. However, the key question is whether such a body can keep pace with the speed of technological development and whether it will become a bureaucratic drag on innovation. The market for independent auditors looks promising but will require the highest level of expertise and trust.