Google, OpenAI and Anthropic are reportedly finalising plans for an independent safety standards body for frontier AI, according to a report by The Information.

The body, tentatively called the Standards Authority for Frontier AI (SAFA), could launch in late 2026 or early 2027. It would support independent pre-deployment testing of advanced AI models, set standards for reporting safety incidents, and define qualifications for AI auditors. The model is said to be inspired by FINRA, the self-regulatory body for US securities.

The labs originally wanted a public-private partnership with government oversight. Those plans stalled after a draft White House executive order failed to win backing, and the companies are now moving ahead with an industry-only body.

The push follows a July proposal by DeepMind chief Demis Hassabis for a FINRA-like AI regulator.

But SAFA already faces criticism. Nvidia CEO Jensen Huang and Meta CEO Mark Zuckerberg oppose the approach, while Cohere CEO Aidan Gomez called it ‘a cartel by any other name’. The three companies have reportedly not reached agreement on whether SAFA should run its own model evaluations.

All reporting traces back to one original story, which the labs have not confirmed, so the body remains a reported proposal.