Microsoft warns AI agents resemble insider threats as frontier models can’t be trusted
Read the original at techcrunch.com→Microsoft wants enterprises to closely watch the agents employees use with systems not controlled by AI labs like OpenAI and Anthropic.
Original headline: "Microsoft says AI agents are like insider threats as ‘frontier models can’t be trusted’"
Coverage timeline
- Oct 19, 22:04 UTC TechCrunch AI lead source Microsoft says AI agents are like insider threats as ‘frontier models can’t be trusted’