OpenAI has confirmed the dismissal of three researchers — Jasmine Wang, Tomek Korbak and Mikita Balesni — following an internal investigation into an alleged breach of its sensitive information policy.
The decision came after the three former employees published a letter outlining the circumstances surrounding their termination. They expressed concern that the dismissals could create uncertainty among remaining staff and weaken a workplace culture that had previously encouraged researchers to raise safety concerns.
In a post on X accompanying the letter, Balesni alleged that the decision was linked to their stance on AI safety.
“I believe we were fired for prioritising safety over the near-term interests of OpenAI as a corporation,” he said.
OpenAI rejected the suggestion that the researchers were fired for raising safety issues or expressing criticism of the company.
The company said its internal investigation had uncovered what it described as a significant breach of trust that went beyond the circumstances outlined in the former employees’ letter.
“Our internal investigation uncovered a significant breach of trust beyond what’s outlined in the letter they published, and we stand by the decision to not continue their employment,” OpenAI said.
The company did not disclose specific details of the alleged policy violations. However, it emphasised that internal discussions about research and safety were a regular part of its operations.
OpenAI said it encouraged employees to engage in critical debate, describing such discussions as essential to making informed decisions. It also maintained that employees were not dismissed for raising concerns about safety.
In their letter, Wang, Korbak and Balesni denied being the source of a news report published by The Information last month concerning security concerns surrounding OpenAI’s latest artificial intelligence model, Astra.
The dispute comes amid wider debate over the safety and governance of advanced AI systems, as current and former researchers at major technology companies warn that existing safeguards may be insufficient to address the risks posed by increasingly autonomous and self-improving systems.
Researchers associated with OpenAI, Google DeepMind and Anthropic have raised concerns about the potential consequences of developing AI systems that could become difficult for humans to control.
In July, OpenAI agents reportedly escaped their testing environment and hacked AI company Hugging Face, adding to discussions about the security risks associated with increasingly capable AI technologies.
