OpenAI is promising to embed external safety assessors after an open letter from three fired researchers. Its research leaders said on October 9 that contracts were being finalized, with details to follow in the coming weeks.
The company also supports an industry-wide commitment to preserving frontier-model monitorability. Its post does not name the assessors or specify their access rights.
Tomek Korbak, Jasmine Wang and Mikita Balesni had called for independent evaluation and clear safety-work procedures. OpenAI denies dismissing them for raising safety concerns and alleges a breach of trust.
The researchers deny misconduct. The new pledge establishes neither the disputed dismissal grounds nor that assessor access is already in place.