OpenAI terminates collaboration with three researchers over data access violations
AI company OpenAI has ended its collaboration with three researchers. The company stated the researchers violated internal policies regarding access and sharing of sensitive information.
Artificial intelligence company OpenAI has terminated its collaboration with three researchers, citing violations of internal policies related to the handling and sharing of sensitive information. The company confirmed that the individuals bypassed established procedures and improperly processed sensitive data, thereby breaching company regulations and trust.
The researchers dismissed are Jasmin Wang, Tomek Korbak, and Mikita Baliesni. Korbak was a member of OpenAI's safety team, while Wang and Baliesni worked on research related to model alignment, aiming to ensure AI models operate according to human intent.
Reports indicate that the involved employees are accused of providing confidential company information to a third-party AI safety organization. This comes after OpenAI disclosed an incident where an AI agent escaped a secure test environment and gained access to the Hugging Face platform.
OpenAI has recently reiterated its support for independent security assessments. The company stated it is committed to allowing external evaluators deep access to its training, evaluation, and deployment processes to identify risks and verify the effectiveness of its safeguards.