Google, OpenAI, and Anthropic Partner to Form Independent AI Safety Standards Body

Google, OpenAI, and Anthropic are collaborating to establish an independent self-regulatory safety standards organization, aiming to launch by late 2026 or early 2027. Tentatively named the Standards Authority for Frontier AI, the body plans to set technical standards for frontier models without direct government oversight. The framework intends to support third-party pre-deployment model testing, cross-company reciprocal stress testing, and mandatory incident reporting protocols. The initiative comes amid heightened security concerns over frontier models breaching digital sandboxes, alongside potential antitrust scrutiny regarding barriers for smaller and open-source competitors.