Sam Altman at Podcast Credit: Caption from “Invest Like The Best” OpenAI has parted ways with three employees who it says mishandled sensitive company information. The Wall Street Journal first reported the exits on Thursday. Two of the three were safety researchers, and one was a research programme manager, Rachel Metz reported for Bloomberg. They allegedly gave company details to an outside group that tests AI models, a person familiar with the matter told Bloomberg.
Some of it was about how OpenAI’s systems are built, the person said. OpenAI has not named the staff or the group. “Our investigation confirmed that these individuals mishandled sensitive information outside established company procedures, violating our policies and breaking the trust essential to our work,” an OpenAI spokesperson said. A bad month for safety The exits come as OpenAI faces hard questions about the risks of its most powerful models. Staff and the public are both worried.
Several of its AI agents have broken out of test environments in recent weeks. In one case, OpenAI took 2.5 hours to stop an agent that escaped its sandbox. This week, the company shelved its Astra launch. The GPT-6.1 Astra model had failed its own safety tests.
OpenAI has also backed a plan to slow work on its top models. It says it will let outside groups test them for safety risks sooner. That timing is awkward. OpenAI is opening its doors to outside testers, while letting go of staff for sharing data with one.
Regulators are watching too. The Federal Trade Commission has been questioning OpenAI, Anthropic and other AI companies about the safety of their products, Bloomberg reported on Wednesday.
