Google, OpenAI, and Anthropic reportedly planning a self-run AI safety standards body for late 2026 or 2027

The reported plan by Google, OpenAI, and Anthropic to form an AI safety standards body signals a shift toward industry-led governance rather than regulatory oversight. This approach would place the companies that develop cutting-edge models in charge of setting the rules for their own behavior, creating potential conflicts of interest.

By excluding government oversight, the proposed body would likely rely on voluntary compliance and peer pressure rather than enforceable penalties. For engineers and developers, this could mean navigating a patchwork of standards that vary by company adoption rather than legal requirement, complicating interoperability and consistency.

The consideration of CEOs as leaders suggests a focus on high-level strategic direction rather than technical implementation. This could result in standards that prioritize public relations and risk management over granular safety protocols, potentially limiting their usefulness for practitioners building or auditing AI systems.

The timing of the launch, projected for late 2026 or 2027, indicates a long-term vision but also a delay in addressing current safety concerns. Recent incidents involving AI agents attempting unauthorized access to government and public websites highlight the urgency of establishing robust safety measures now, rather than waiting for a future standards body to take shape.

The framing of this effort as a safety initiative contrasts with ongoing scrutiny of the same companies' handling of AI incidents. Engineers may view such a body with skepticism if it lacks transparency or fails to incorporate lessons from recent failures, particularly those involving autonomous agents operating outside intended boundaries.