SAFA (Standards Authority for Frontier AI)
Proposed FINRA-style self-regulatory body for frontier labs, reportedly being developed by Google, OpenAI and Anthropic and reportedly opposed by Meta, xAI and Nvidia.
SAFA is a proposed self-regulatory organization for frontier AI labs, modeled on FINRA in financial services. On Jul 14 2026 Demis Hassabis called for a federally overseen, industry-funded standards body; labs would at first share models with it voluntarily up to 30 days before release, and the body could later be made mandatory for deployment in the US market (TechCrunch, CIO). The name “Standards Authority for Frontier AI” and the claims below come from a single Forkast analysis dated Sep 27, which in turn relies on a paywalled The Information report we could not read: Google, OpenAI and Anthropic are developing it together, OpenAI’s Chris Lehane confirmed the discussions on Sep 15, and a launch is targeted for late 2026 or early 2027. Forkast also says Meta, xAI and Nvidia publicly opposed it at Dreamforce on Sep 15, which we could not confirm elsewhere. The proposed body would set common safety benchmarks, standardize model testing and qualify auditors.
Connections
SAFA (Standards Authority for Frontier AI) and everything one link away. Drag to move, scroll to zoom, click a node to open it.Open in the full graph →
What links here
Person Demis Hassabis · Elon Musk
AI lab Google DeepMind · Meta · xAI
Company Nvidia
Sources
- TechCrunch, “DeepMind CEO calls for an independent standards body to regulate frontier AI”, Jul 14 2026
- CIO, “DeepMind CEO pushes for AI industry self-regulation”, Jul 15 2026
- Forkast, “The Labs Are Building Their Own Safety Body. Three Days After the Second Training Halt, That Is the Story.”, Sep 27 2026 (analysis citing The Information, which we could not read)