Microsoft says AI agents are like insider threats as ‘frontier models can’t be trusted’
Microsoft wants enterprises to closely watch the agents that employees use with systems not controlled by AI labs like OpenAI and Anthropic.
Why it matters
Microsoft wants enterprises to closely watch the agents that employees use with systems not controlled by AI labs like OpenAI and Anthropic Reported by TechCrunch AI on October 19, 2026.