OpenAI Backs House Bill on Safety Evaluations

The notable element is less the endorsement itself than the direction of travel: a leading lab publicly aligning with formal evaluation requirements rather than resisting them. That shifts the policy conversation from whether assessments should exist to how they are scoped and who administers them — the details left unstated here. For model developers and enterprise buyers, the practical question is whether voluntary backing translates into binding, standardized testing across the field, or remains a single-company gesture. Whether other major labs follow, and whether the bill's undefined scope narrows or broadens, is the open question.