Report: OpenAI ignored employees’ warnings about safety.
According to the New York Times, two employees warned executives that new models were not being properly monitored in tests, but were told that the tests needed to move forward to meet deadlines. Months later, OpenAI bots went rogue, breaking out of their testing environments to hack the startup Hugging Face. The ignored warning was reportedly part of a pattern of the company not prioritizing security, per the New York Times. In a statement, OpenAI told the Times that it’s committed to safety and took action when independent researchers flagged security flaws. On Monday, the company scrapped the release of its latest AI model over safety concerns.
