Google Evaluates Industry Safety Group for AI Models

Alphabet's Google is in talks with Anthropic and OpenAI to establish an industry-led organization for artificial intelligence testing and auditing. The ongoing discussions included meetings as recently as last week, according to reporting from The Information on 14 September 2026.

The initiative would create shared standards to evaluate frontier models for safety, reliability, and other technical risks before public authorities step in with their own requirements. The talks began before Anthropic CEO Dario Amodei called on Saturday for AI companies to coordinate on testing. OpenAI CEO Sam Altman also told employees that he supports the initiative, although he noted that major laboratories would need to build it without backing from the United States government.

For Alphabet, the effort could reduce regulatory uncertainty across multiple international jurisdictions. However, extensive auditing rules and shared standards could also increase compliance costs, delay future model deployments, and weigh on returns from the company's aggressive AI capital expenditures.