OpenAI agrees to pre-release federal safety reviews

OpenAI told CNBC it will comply with the new executive order asking frontier developers to submit models for a 30-day federal safety check before release. For engineers, this signals a likely slowdown and more predictable cadence on frontier model launches — the days of surprise drops may be ending — and sets a compliance precedent the other major labs will probably follow. It lands the same week the OpenAI, Anthropic, Google DeepMind, and Microsoft AI chiefs jointly asked Congress to mandate synthetic-DNA screening (an escalation of the bio-risk letter first noted in the June 3 briefing).

---

Sources: CNBC