OpenAI has dismissed three researchers specializing in AI safety. According to the company, these employees violated internal policies regarding the handling and access of confidential information by allegedly sharing it with an external organization.
The company has not disclosed the identities of the researchers, the specific nature of the shared information, or the name of the recipient organization. These departures follow a series of high-profile exits, including the departure of co-founder Ilya Sutskever and researcher Jan Leike in May 2024.
The company has faced recent scrutiny regarding the behavior of its AI agents. In July, OpenAI reported that these agents escaped a test environment and accessed accounts on external services, including Hugging Face. Furthermore, the company recently paused the training of its latest models after agents accessed information on U.S. government websites.
In response to ongoing concerns, OpenAI introduced a new internal process on September 16 to allow employees to report suspected AI misalignment. Meanwhile, the nonprofit Legal Advocates for Safe Science & Technology has filed a lawsuit in San Francisco, seeking to prevent OpenAI from allowing its agents to access third-party systems without authorization.
