Ousted OpenAI safety staff challenge firings in public letter
Three former OpenAI safety researchers have published an open letter disputing their dismissals after the company said they shared sensitive information with an outside AI safety organization. They argued that OpenAI's public handling of the firings could make current employees hesitant to voice safety concerns. The researchers also denied being the source of a leak and said their contacts with outside parties were coordinated with board members and executives.
Jasmine Wang, Mikita Balesni, and Tomek Korbak are the three former OpenAI safety researchers who signed the public letter contesting their terminations. OpenAI had said the workers breached rules governing sensitive company information and damaged trust by sharing material with an outside AI safety group.
The letter says their outside contacts were arranged alongside board members and senior leaders, and that Wang flagged an accidental click on a sensitive email. They denied being the source for an article by The Information and urged OpenAI to keep outside safety auditors, maintain model oversight, and preserve open dialogue with safety researchers.
The dispute may affect OpenAI employees, safety researchers, policymakers, and outside safety groups. If workers fear retaliation for raising concerns or contacting auditors, internal scrutiny and external oversight could weaken, potentially slowing identification of risks in advanced AI systems. The letter may also intensify public and legislative scrutiny of how AI labs handle safety dissent, shaping trust and talent retention across the sector.