Google, OpenAI and Anthropic move closer to AI safety standards body

- Google,
OpenAI and
Anthropic are reportedly planning a new industry venue, tentatively called the Standards Authority for Frontier AI, with a possible 2026–27 launch.
- The three companies tentatively call the organization the Standards Authority for Frontier AI (SAFA), and the plan is for it to operate independently and address gaps in government oversight, according to
The Information.
- The proposed body would sit alongside the Frontier Model Forum, founded in 2023, but the proposal has not clarified whether it would publish guidance, run evaluations or require members to act on failures.
- OpenAI and Anthropic have also discussed an agreement allowing each company to test the other's commercially available AI models for vulnerabilities and unexpected behaviour, while restricting the retention of data obtained during those evaluations.
- OpenAI has said it supports working with other frontier labs on voluntary standards and also advocates mandatory U.S. safety requirements.
Source details
- Latest report
- 24 Sep 2026 11:43 UTC
- First seen here
- 24 Sep 2026 13:46 UTC
- Outlets
- 3