Overview
- Multiple outlets reported on Friday that Google, OpenAI and Anthropic are finalizing plans for a provisional Standards Authority for Frontier AI, often called SAFA, with a possible launch in late 2026 or early 2027.
- Organizers have reportedly approached former White House AI adviser Sriram Krishnan and ex-OSTP director Arati Prabhakar about leading the group and have discussed senior figures for the board.
- OpenAI confirmed to CNBC that it has held active talks with Google and Anthropic about a shared safety framework, which sources say helped move discussions from concept to concrete organizing.
- A key unresolved decision is whether SAFA will only set standards or also run direct, in-house model assessments because federal capacity for deep technical inspections is seen as limited.
- Commentators warn that a self-regulatory body run by the largest labs could raise compliance costs, create barriers for smaller developers and shape future law and procurement standards.