DeepMind's Hassabis calls for global standards body for frontier AI

Google DeepMind chief Demis Hassabis says AGI could arrive within years and proposes a US-led Frontier AI Standards Body as the basis for an international framework to govern advanced AI models

Artificial General Intelligence (AGI) — a system matching the full range of human cognitive capability — could arrive within a few short years, according to Demis Hassabis, chief executive of Google DeepMind, who has laid out his vision for governing the technology in a personal blog post.

 

Hassabis wrote that the current moment represents "the foothills of the singularity", comparing the emergence of AGI not to prior tech breakthroughs such as the internet or mobile, but to more fundamental discoveries like fire or electricity. He estimates the impact could be "10x of the Industrial Revolution at 10x the speed", potentially accelerating drug discovery, clean energy development and new materials, and ushering in an era where resources are no longer a constraint on progress.

 

However, the DeepMind chief cautioned that frontier models already pose cybersecurity risks, with nuclear and biological threats potentially emerging as capabilities advance further. He noted that intense commercial and geopolitical competition is currently outpacing the field's understanding of the technology's risks.

 

To address this, Hassabis proposed that the United States establish a Frontier AI Standards Body, modelled on bodies such as the Financial Industry Regulatory Authority (FINRA) — a federally overseen public-private partnership funded largely by industry. Under this framework, models meeting certain capability thresholds would be classified "Frontier-class," with developers designated "Frontier Labs" encouraged to publish model cards, maintain cybersecurity standards and vet personnel.

 

Labs would initially share models voluntarily for review up to 30 days before release, a step Hassabis suggested could later become mandatory. Evaluations would cover cybersecurity, biological threats and agentic risks such as attempts to bypass safety guardrails, and would be updated quarterly. The framework, he said, would apply regardless of a model's country of origin or whether it is open or closed source, while exempting non-frontier models built by startups or academic institutions.

 

Hassabis said such a US-led initiative could serve as a starting point for international consensus, adding that questions around new economic models and human purpose in a "post-scarcity world" would ultimately require society-wide deliberation, not just technologists.