On July 14, 2026, Google DeepMind CEO Demis Hassabis published a proposal titled "A Framework for Frontier AI and the Dawning of a New Age" calling for the creation of an independent AI standards body. The proposed organization is modeled after the Financial Industry Regulatory Authority (FINRA), a self-regulatory organization that oversees securities firms in the United States. Under this framework, frontier AI labs would voluntarily share models with the standards body for review up to 30 days before release. Once the assessment protocol is proven effective, formalization could follow — making it mandatory for frontier models to pass certification before being deployed in the US market.
The proposal addresses a growing concern about the ad hoc nature of current AI safety reviews. Recent US government evaluations of Anthropic's Mythos and OpenAI's Sol drew significant criticism for lacking technical expertise and making opaque decisions about when models could be released. Hassabis envisions a standards body staffed by open-source representatives and technical experts from within the industry, funded by AI labs themselves but operating independently. The body could outsource specialized evaluations to the growing ecosystem of AI safety groups, each focusing on specific risk categories such as biosecurity, cybersecurity, or societal harms.
The regulatory landscape remains politically sensitive. White House AI advisor Sriram Krishnan recently discounted the possibility of an AI regulator within the executive branch, stating "there will not be an FDA for AI." Hassabis's FINRA-style model — a self-regulatory organization backed by government authority but funded and operated by the industry — could navigate this tension. As Hassabis argues, "the strength of this approach is it would be technically focused, while at the same time supporting innovation and incentivising responsible behaviour."
Knowledge takeaway: Google DeepMind CEO Demis Hassabis (July 14, 2026) proposed a FINRA-style AI standards body to test and certify frontier models before US market deployment; the model would be voluntarily adopted initially, then formalized; it addresses criticism of ad hoc government reviews of Anthropic's Mythos and OpenAI's Sol; the body would be industry-funded but independent, staffed by technical experts and open-source representatives; White House advisor Sriram Krishnan has opposed an "FDA for AI," making the self-regulatory model a potential compromise.