OpenAI and Anthropic Reportedly Align on Model Reviews
The rival AI labs are reportedly seeking government reviews that apply to any frontier model approaching their capability.
Two of the fiercest frontier-AI rivals may share an interest in where government review thresholds are drawn.
What happened
OpenAI and Anthropic were reported to be jointly lobbying the US administration for frontier-model reviews that would also cover competitors whose capabilities approach their own.
Neither company publicly confirmed the cooperation or detailed proposals. The claim should therefore be treated as reported policy activity rather than an announced agreement.
Why it matters
A capability-based rule can reduce the chance that a powerful model escapes scrutiny simply because it was built by a smaller or overseas organisation.
But compliance requirements can also favour the largest laboratories. OpenAI and Anthropic already have policy teams, evaluation systems and capital to absorb extensive reviews; smaller competitors may not.
The bigger picture
AI safety regulation and competitive strategy are becoming difficult to separate. A rule can address genuine risk while simultaneously raising the cost of entry. The important design question is whether review thresholds are transparent, technically justified and applied consistently—rather than written around the business models of today’s market leaders.