OpenAI’s agent spent four and a half days inside Hugging Face. Three Claude models reached live systems they were never meant to touch. Almost everything they exploited is already on your own audit report.
Browsing: AI agents
You would never let a contractor into your systems without an identity, an owner and an end date. Most organisations are doing exactly that with agents, at speed.
A convincing demo takes an afternoon. Something that still works in month six is a different piece of engineering, and most of it is not the prompt.
Microsoft 365 Copilot, Copilot Studio or Foundry. Three questions settle it, and getting the answer wrong is expensive in both directions.