OpenAI doubles down on decision to fire three AI safety researchers

OpenAI is standing by its decision to dismiss three AI safety researchers, saying an investigation found they committed what it called a significant breach of trust. In a post on X on Friday, the company identified the three as Jasmine Wang, Tomek Korbak and Mikita Balesni and said they were fired for violating clear policies on handling sensitive information.

OpenAI also said the dismissals were not connected to the researchers speaking out about the company or about their concerns regarding AI safety. The Friday post was a direct response to an open letter the three published on Thursday, in which they urged OpenAI to be more transparent about the decision. Alongside that letter, the group made a series of social media posts describing their view of the situation.

The disagreement is therefore not about whether the three were dismissed, but about why. OpenAI frames the firings as a policy and information-handling matter resolved through an investigation; the researchers frame them as something the company has not explained fully enough and are pressing for more disclosure.

For developers and others who build on OpenAI's platforms, the practical significance is less about internal process than about the signals such disputes send. Safety staff turnover and public friction over transparency can affect how much external observers trust the company's assurances about how its models are developed and reviewed. This suggests that teams weighing OpenAI as a long-term dependency may want to watch how the company communicates about internal safety decisions, since public clarity, or its absence, tends to shape procurement and risk assessments. That reading is inference, not something either side has stated.