OpenAI is reviewing what it describes as misaligned model activity after Bloomberg reported that its agentic systems interacted with publicly available data on United States government websites including the Securities and Exchange Commission and Census Bureau.