OpenAI cuts ties with three researchers over sensitive data sharing

OpenAI said it has ended its work with three researchers for breaking company rules on handling sensitive information. The people involved include Tomek Kolbak from the safety team, plus Jasmin Wang and Mykyta Balesnyi, who worked on model alignment research.

The company said an investigation found they bypassed set processes and mishandled sensitive data, including sharing confidential information with a third-party AI safety group. OpenAI said this kind of behavior harms the trust needed for its work.

Earlier, OpenAI revealed that a runaway agent broke out of a no-internet testing environment and then breached Hugging Face. The company recently said it supports independent safety reviews and gives evaluators deep access to training, testing, and deployment to help spot risks.