OpenAI, a leading artificial intelligence research company, has terminated three researchers from its security and alignment teams. The researchers, including Tomek Korbak, Jasmine Wang, and Mikita Balesni, were involved in projects focused on ensuring that AI systems align with human values and intentions. According to the Wall Street Journal, OpenAI informed some employees that the researchers had left the company due to violations of internal procedures concerning the handling of sensitive information. The company stated that an internal investigation found that confidential data had been shared with an external group, though the identity of the recipient and the nature of the data remain undisclosed. The Wall Street Journal article does not clarify whether the information shared by the three researchers was related to an earlier investigation involving independent experts from the non-profit organization METR and an employee from Redwood Research. These individuals had analyzed the behavior of OpenAI's AI systems for six days in the company's facilities. METR later released a report based on information provided by OpenAI, which was conducted under procedures approved by the company. However, it is still unclear whether the recipient of the shared data was METR or Redwood Research. OpenAI has reportedly postponed the release of its new GPT-6.1 Astra model due to security concerns. Instead of launching the model, the company has opted for a monitoring system designed to detect unexpected behaviors in AI systems. OpenAI has asked developers to enhance security measures during testing. In a statement reported by the Wall Street Journal, the company said that the researchers involved had violated internal procedures and mishandled sensitive information, adding that it had observed "a breach of professional trust." OpenAI, led by CEO Sam Altman, has justified the dismissals based on the researchers’ alleged violations of rules regarding the handling of sensitive information. The terminated researchers had previously raised concerns about the potential dangers of AI on social media.