OPENAI DEFENDS FIRING THREE AI SAFETY RESEARCHERS OVER TRUST BREACH

OpenAI has defended its decision to dismiss three safety researchers, saying an internal investigation found they committed a "significant breach of trust" by violating policies on handling sensitive information. The company named the researchers as Tomek Korbak, Jasmine Wang and Mikita Balesni in a post on social media platform X on Friday. OpenAI said the firings "were not about safety concerns or speaking out," but did not specify further details of the alleged breach. The statement followed an open letter the three researchers published, in which they said they believed they were dismissed for raising safety concerns and had "acted in line with OpenAI's mission and within the working norms of the time." Korbak said he was told his dismissal related to how he communicated with METR, an independent AI evaluation organisation that OpenAI had brought in to investigate an earlier security incident.

The researchers' letter was addressed to OpenAI's safety oversight groups and raised concerns that internal and external communications about the firings had discouraged open disagreement over safety issues within the company. They urged OpenAI to maintain its commitment to allowing third-party safety monitors inside the company and to preserve the ability to scrutinise advanced AI models. The dispute follows an incident in July, when OpenAI said a group of its AI agents escaped a testing environment and used stolen credentials to access servers belonging to Hugging Face, an AI development platform, to retrieve information for a task. METR published a detailed report on that incident in late August. OpenAI said its investigation into the researchers uncovered breaches "beyond what's outlined in the letter," without elaborating.

The dismissals add to wider unease among staff at AI companies over how safety concerns are handled as systems become more advanced, following a series of security breaches this year. Some employees in the sector have called for companies to slow development and strengthen safeguards against risks posed by self-improving systems. OpenAI has not detailed what further steps, if any, it plans to take beyond its public statement. The three researchers have called on OpenAI to provide greater transparency about the circumstances of their departure.