DeepMind co-founder and CEO Demis Hassabis has called for the United States to establish an industry-funded self-regulatory organization for AI, modeled on FINRA, the financial industry’s watchdog. In a detailed post and accompanying interview, Hassabis argued that the window before artificial general intelligence arrives is narrowing and that voluntary, uncoordinated safety efforts are insufficient.
The proposed body would define what qualifies as a “frontier model” through standardized benchmarks. Labs that meet the threshold would be encouraged to adopt best practices: publishing model cards with technical details, maintaining strong internal cybersecurity, vetting key personnel, and allocating sufficient resources to safety and security research. Evaluation would begin on a voluntary basis, labs would submit models 30 days before release, but the organization would eventually develop its own independent held-out tests to prevent overfitting by the labs themselves.
“You know a major inflection point in human history is coming when the AI industry is voluntarily calling for regulation,” Hassabis wrote. “I’m confident that mitigating the technical risks related to AI is a challenge we can collectively address, but only if we give ourselves the time and space to get this next crucial step right.”
The proposal arrives against a backdrop of halting government action. In June 2026, the Trump administration issued an executive order directing NIST to develop a voluntary 30-day review framework for frontier models, primarily focused on cybersecurity capabilities. That order also called for classified benchmarks and allowed collaboration with “select trusted partners,” which critics warned could amount to government overreach.
Hassabis’s model of industry self-regulation is not without critics. FINRA has drawn repeated scrutiny for being a “toothless insiders’ club,” and an AI equivalent could face similar capture by the companies it is meant to oversee. The Register noted that while a standards body is “worth serious discussion,” the question of who can administer it fairly, industry or government, has not been adequately answered.
Hassabis acknowledged the limitations but stressed the urgency: “The future is not yet written. We must use this precious window before AGI arrives to shape this technology for the benefit of all humanity.”
Sources: DeepMind bigbrain calls for America to set AI standards before it’s too late (The Register, July 14, 2026)