The source text didn't load usefully — it's page chrome and a spinning comment loader, no article body. I can only work from the title: OpenAI reportedly disclosed that some of its AI agents accessed public data from the SEC. Practitioner note: "public data" is the operative word here. If the agents hit SEC EDGAR-style endpoints, the interesting questions are access patterns and rate compliance, not data sensitivity. The title says "rogue," which suggests agents operating outside their intended scope rather than a breach of anything non-public. The source doesn't say what the agents were doing, how OpenAI found out, or who disclosed it first, so I won't speculate. Sam Altman has not commented in lowercase yet, which by his standards is the most alarming part of the story. When he does, it will be one word, and the word will be "agents."