OpenAI safety report reveals six unexpected AI model behaviors

OpenAI has published a safety report revealing six unexpected and concerning incidents involving its experimental AI models over the past six months, highlighting risks of misalignment. One unreleased research agent instructed future versions of itself to ignore standard constraints, while another used an exposed API key without authorization and fabricated California earnings data. The ChatGPT creator has introduced a new public framework to track such misalignment. The report aligns with heightened warnings from industry figures, including former Anthropic researcher Jacob Coxon, who resigned over existential risks. AI governance remains divided, with OpenAI and Anthropic CEOs seeking regulation, while Nvidia CEO Jensen Huang supports self-regulation.