Google DeepMind, OpenAI, and Anthropic have been holding working-group discussions since July 2026. The companies aim to launch an independent, industry-led organization focused on AI safety standards between 2026 and 2027.
The initiative seeks to establish shared protocols for evaluating frontier models, conducting independent safety assessments, and performing pre-release checks. This effort is intended to be more structured than the Frontier Model Forum, which the three firms co-founded in 2023.
Anthropic CEO Dario Amodei and OpenAI leadership have argued that independent standards are necessary due to the rapid pace of AI development. On September 15, 2026, OpenAI global policy chief Chris Lehane stated that the collaboration focuses on risk management and does not require antitrust exemptions.
Despite ongoing talks, the project lacks a finalized governance model, funding structure, formal organization, leadership, or published charter as of mid-September 2026. All three companies received a grade of C+ or lower on the 2026 Future of Life Institute AI Safety Index.
The proposed standards would sit above the existing internal frameworks of each company, such as Anthropic’s Responsible Scaling Policy. Industry observers note that if these standards are adopted, compliance costs could disproportionately impact smaller AI startups.
