OpenAI Fires Three Safety Researchers

Substack · Watch/Read · (gathered 2026-10-14)

• Zvi Mowshowitz argues that OpenAI’s firing of three prominent safety researchers (Tomek Korbak, Mikita Balesni, Jasmine Wang) for vague reasons is a “terrible look” and an “own-goal” that severely damages trust and creates a “chilling effect” on safety advocacy.
• The researchers claim they were fired for safety advocacy, receiving different, pretextual reasons, while OpenAI vaguely cites “mishandling sensitive company information” and a “significant breach of trust” without providing details, leading to “earned skepticism” given OpenAI’s track record.
• The dismissed researchers were central to critical safety efforts like “pacing the frontier,” the METR investigation, and monitorability, making their loss a significant blow to the safety ecosystem and undermining OpenAI’s stated commitments to external auditors and open dialogue.
• The lack of transparency and reliance on “non-explanations” ensures that “no one is buying” OpenAI’s narrative, fostering a perception that safety advocacy is being suppressed and making it harder for other employees to speak up.