July 15, 2026, (Inside AI) — Demis Hassabis, CEO and founder of Google DeepMind, has called for the creation of an independent US regulatory body to set standards for frontier AI labs. In a detailed blog post, the Nobel laureate proposed a public-private partnership modeled on the Financial Industry Regulatory Authority (FINRA) to assess the most advanced AI models before public release.
Hassabis's proposal arrives as the US government intensifies scrutiny of frontier models, including Anthropic's Mythos and OpenAI's GPT-5.6, sparking debate over pre-release evaluation. He recommends a 30-day assessment window—longer than current voluntary evaluations but shorter than many formal regulatory reviews.
In his post, 'A Framework for Frontier AI and the Dawning of a New Age,' Hassabis argues that artificial general intelligence (AGI) could arrive within years, making common safety standards urgent. He writes:
"When we look back on this time in the decades to come, I think we will realise we were standing in the foothills of the singularity - nothing less than the dawning of a new age for humanity."
Hassabis likens AGI's potential impact to electricity and fire, predicting it could accelerate breakthroughs in drug discovery, clean energy, and materials while driving massive economic growth. He states:
"The magnitude of this technology's impact will be unprecedented, perhaps 10x that of the Industrial Revolution at 10x the speed."
Yet he warns that AI capabilities are outpacing risk comprehension, highlighting cybersecurity, biological threats, and future autonomous systems. He notes:
"Nobody in the world knows for sure what is going to happen from here, and even the experts disagree."
Hassabis advocates cautious optimism, urging policy that balances innovation with responsibility and international collaboration. His proposed standards body would define technical benchmarks for frontier models, conduct pre-deployment evaluations, and test for risks including deception and safety bypass attempts.
Initially, developers would voluntarily submit models 30 days before launch. If effective, assessments could become mandatory for US releases. Frontier Labs would also publish model cards, strengthen cybersecurity, invest in safety research, and cooperate on post-release fixes.
Hassabis envisions the framework evolving with AI, with regularly updated benchmarks and independent expert input. Though US-focused, he says it could underpin international standards. He concludes:
"We must use this precious window before AGI arrives to shape this technology for the benefit of all humanity."
The proposal echoes FINRA's self-regulatory model, where industry and government collaborate on oversight. Critics note that voluntary schemes may lack teeth, but Hassabis's phased approach—starting voluntary, then potentially mandatory—aims to build trust and data for eventual regulation. This mirrors the trajectory of financial regulation post-2008 crisis.
Hassabis's timeline for AGI aligns with other industry voices, including OpenAI CEO Sam Altman, who has suggested AGI could emerge this decade. However, definitions of AGI remain contested, with some experts arguing current systems lack true understanding. The proposed body would need to precisely define frontier models—a challenge given rapid progress.
Hassabis also stresses that safe AGI development could usher in "a new golden age of scientific discovery and progress." His dual emphasis on promise and peril reflects DeepMind's long-standing focus on safety, including its early work on AI alignment and the establishment of an internal safety research group.
The blog post comes as governments worldwide grapple with AI governance. The EU's AI Act takes a risk-based approach, while the US has relied on executive orders and voluntary commitments. Hassabis's proposal could influence upcoming US legislation, though political hurdles remain.
By invoking the singularity and historical transformations, Hassabis frames AGI as both an existential opportunity and a test of collective foresight. His call for a standards body is less a technical fix than a bet on institutional design to steer a technology he believes will reshape civilization.
