Microsoft says AI agents are like insider threats as ‘frontier models can’t be trusted’
companiesTechCrunch AIOct 19, 2026
Microsoft wants enterprises to closely watch the agents that employees use with systems not controlled by AI labs like OpenAI and Anthropic.
Impact: LowConfidence: MediumNovelty: Incremental
companiestechcrunch-aiagents