Largest AI Labs Propose Joint Safety Standards Initiative

A joint AI standards organization is being proposed by Anthropic, OpenAI, and Google, according to a report by The Information. This would enable the largest AI labs to establish safety standards for their own models.

The introduction of common standards could increase certainty among companies evaluating whether to use frontier models for sensitive tasks or not. However, this framework might also empower the largest players in the market, making it even more challenging for smaller laboratories to enter the AI space.

Demis Hassabis, head of Google DeepMind, suggested an initiative for a U.S.-based Frontier AI Standards Body inspired by FINRA. The proposal includes voluntary submission of advanced technologies for assessment with a 30-day evaluation period before release.