Blab AI
DeepMind CEO Proposes FINRA-Style Self-Regulatory Body for Frontier AI Models
On July 14, 2026, DeepMind chief executive Demis Hassabis sparked a new debate by announcing on social media that the company would set up an independent, FINRA‑style self‑regulatory body to vet the most powerful artificial‑intelligence models before they reach the market.
The idea has quickly gathered momentum across the tech ecosystem. Microsoft AI CEO Mustafa Suleyman and corporate chief Satya Nadella both backed the proposal, while Block founder Jack Dorsey, Box chief Aaron Levie, OpenAI founder Sam Altman, and co‑founder Elon Musk all praised the framework as "thoughtful." Former Trump administration AI czar David Sacks also endorsed the plan, arguing it was preferable to direct government regulation.
Bloomberg reported that the Trump administration is reviewing the proposal. Treasury Secretary Scott Bessent, according to unnamed sources familiar with the discussions, helped develop the idea, and the plan is under review by White House Chief of Staff Susie Wiles. The proposed body would be overseen by the Securities and Exchange Commission (SEC), which has statutory authority to delegate oversight to self‑regulatory organizations.
Hassabis outlined the body’s operations. Funding would come from leading AI labs, and the board would feature independent technical experts, representatives from the open‑source AI community, and vendor representatives. The organization would develop assessment protocols for frontier AI, including capability benchmarks to identify which models qualify. It would conduct independent safety and security testing in partnership with government agencies on matters that intersect with national security.
Companies producing frontier models would be encouraged to adopt governance standards such as publishing model system cards, following strong cybersecurity protocols, and funding safety research. Initially, the organization would invite voluntary submission of models at least 30 days before release. If the testing proves effective, the process could become mandatory for any model distributed in the United States.
Critics point to potential conflicts of interest. FINRA has faced criticism for prioritizing the interests of brokerage firms that fund it. Senator Elizabeth Warren has repeatedly accused the regulator of protecting its members rather than investors. Whistleblowers have claimed that FINRA has been lax in pursuing problematic brokers. A study by SLCG Economic Consulting found that after FINRA gained the power to label high‑risk brokers as "restricted," it failed to apply the designation to any firm, despite evidence that several should have been labeled.
Industry analysts also question the feasibility of self‑regulation. Nader Henein, a VP analyst at Gartner, told CIO that "self‑regulation is not viable" and that tech vendors lack the capacity to self‑regulate. Carmi Levy, an independent tech analyst, described Hassabis’s proposal as a "self‑serving roadmap for an industry bent on racing to the AI horizon regardless of the harms." Yoshua Bengio, scientific director of the nonprofit AI safety organization LawZero, criticized the voluntary nature of the model and called for a clear roadmap to transition from voluntary to mandatory standards.
The international dimension remains uncertain. If the body’s standards become a requirement for U.S. distribution, foreign vendors would need to comply to reach U.S. users. The standards could, in theory, become a de‑facto global benchmark, similar to how the U.S. Food and Drug Administration’s approvals influence pharmaceutical regulation worldwide. However, the body would not regulate models developed for military or government use outside the commercial distribution chain, leaving a potential gap for high‑risk models.
At present, the proposal is gaining traction among industry leaders and is under consideration by the Trump administration. The SEC’s involvement would give the organization a degree of legitimacy, but questions about funding, governance, and enforcement remain. The next steps will likely involve formal discussions between DeepMind, the Treasury, and the SEC, as well as broader stakeholder consultations.
The outcome of this initiative could shape the regulatory landscape for frontier AI in the United States and potentially influence global standards. Until a formal body is established, the debate over how best to balance innovation with safety and security continues.
The idea has quickly gathered momentum across the tech ecosystem. Microsoft AI CEO Mustafa Suleyman and corporate chief Satya Nadella both backed the proposal, while Block founder Jack Dorsey, Box chief Aaron Levie, OpenAI founder Sam Altman, and co‑founder Elon Musk all praised the framework as "thoughtful." Former Trump administration AI czar David Sacks also endorsed the plan, arguing it was preferable to direct government regulation.
Bloomberg reported that the Trump administration is reviewing the proposal. Treasury Secretary Scott Bessent, according to unnamed sources familiar with the discussions, helped develop the idea, and the plan is under review by White House Chief of Staff Susie Wiles. The proposed body would be overseen by the Securities and Exchange Commission (SEC), which has statutory authority to delegate oversight to self‑regulatory organizations.
Hassabis outlined the body’s operations. Funding would come from leading AI labs, and the board would feature independent technical experts, representatives from the open‑source AI community, and vendor representatives. The organization would develop assessment protocols for frontier AI, including capability benchmarks to identify which models qualify. It would conduct independent safety and security testing in partnership with government agencies on matters that intersect with national security.
Companies producing frontier models would be encouraged to adopt governance standards such as publishing model system cards, following strong cybersecurity protocols, and funding safety research. Initially, the organization would invite voluntary submission of models at least 30 days before release. If the testing proves effective, the process could become mandatory for any model distributed in the United States.
Critics point to potential conflicts of interest. FINRA has faced criticism for prioritizing the interests of brokerage firms that fund it. Senator Elizabeth Warren has repeatedly accused the regulator of protecting its members rather than investors. Whistleblowers have claimed that FINRA has been lax in pursuing problematic brokers. A study by SLCG Economic Consulting found that after FINRA gained the power to label high‑risk brokers as "restricted," it failed to apply the designation to any firm, despite evidence that several should have been labeled.
Industry analysts also question the feasibility of self‑regulation. Nader Henein, a VP analyst at Gartner, told CIO that "self‑regulation is not viable" and that tech vendors lack the capacity to self‑regulate. Carmi Levy, an independent tech analyst, described Hassabis’s proposal as a "self‑serving roadmap for an industry bent on racing to the AI horizon regardless of the harms." Yoshua Bengio, scientific director of the nonprofit AI safety organization LawZero, criticized the voluntary nature of the model and called for a clear roadmap to transition from voluntary to mandatory standards.
The international dimension remains uncertain. If the body’s standards become a requirement for U.S. distribution, foreign vendors would need to comply to reach U.S. users. The standards could, in theory, become a de‑facto global benchmark, similar to how the U.S. Food and Drug Administration’s approvals influence pharmaceutical regulation worldwide. However, the body would not regulate models developed for military or government use outside the commercial distribution chain, leaving a potential gap for high‑risk models.
At present, the proposal is gaining traction among industry leaders and is under consideration by the Trump administration. The SEC’s involvement would give the organization a degree of legitimacy, but questions about funding, governance, and enforcement remain. The next steps will likely involve formal discussions between DeepMind, the Treasury, and the SEC, as well as broader stakeholder consultations.
The outcome of this initiative could shape the regulatory landscape for frontier AI in the United States and potentially influence global standards. Until a formal body is established, the debate over how best to balance innovation with safety and security continues.
