OpenAI's internal coding agent misalignment monitoring (openai.com)

Describes OpenAI's chain-of-thought monitoring approach for detecting misalignment risks in internal coding agents and strengthening AI safety safeguards through real-world deployment analysis.