Microsoft AI agents risk: Guardian Agents launch for firms
Microsoft AI agents risk is now framed as an insider threat: its new Guardian Agents tool monitors the agents firms use, kept apart from model makers such as OpenAI and Anthropic.
Microsoft introduced new security features for enterprises on Monday, led by Guardian Agents, which monitors the AI agents a company uses.
Microsoft says frontier AI models cannot be trusted because they are unpredictable, and it treats agents as possible insider threats.
Its Defender product can now detect up to 60 types of agents that employees bring into the workplace, including Claude Code, Codex and OpenClaw.
Satya Nadella wants firms to run oversight systems that are separate from the AI labs whose agents they deploy.