Breaking

OpenAI, Anthropic and Google Discuss AI Safety Standards Body as Altman and Amodei Back More Testing

September 15, 2026 at 15:01 EDT

Foundation Models
AI Agents
Policy & Regulation

OpenAI, Anthropic and Google have discussed creating an AI industry standards body as leading executives publicly press for more coordinated testing and auditing of frontier models. The talks, described as ongoing, come as OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei endorse stronger third-party oversight for systems that are becoming more capable and more agentic.

September 2026 · Frontier AI governance

AI labs move toward shared safety tests

OpenAI, Anthropic and Google have discussed an industry standards body as Sam Altman and Dario Amodei back more coordinated testing and auditing for frontier AI systems.

labs in standards-body talks

steps in the pacing framework

70%

maximum refusal rate in one evaluation area

The cited risk window

A 6-12 month concern becomes the clearest numeric urgency signal.

6 monthslower bound

12 monthsupper bound

The concern is not a halt in AI work, but whether safeguards, audits and interpretability can catch up before more capable agent swarms arrive.

1 · Embedded evaluators

Outside reviewers gain employee-like access to inspect safety practices and incidents.

2 · Democratic coordination

Frontier labs align on common standards and limits on unchecked progress.

3 · Global coordination

Governments seek wider agreements where verification is difficult but consequential.

Why supporters see progress

Independent checks could make safety claims more verifiable and create common evaluation practices across leading labs.

Why critics are wary

Industry-led coordination could lack enforcement or raise barriers that favor large incumbent AI companies.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.