Microsoft Launches Open Standard to Control AI Agents

Microsoft Launches Open Standard to Control AI Agents
Microsoft has unveiled Agent Control Specification (ACS), an open source standard giving developers a consistent way to govern AI agent behavior across different environments. The spec lets developer, compliance, and security teams define policies dictating what agents can do, what they must avoid, when human approval is needed, and what actions to log. ACS addresses fragmented, hard-to-audit controls that developers currently patch together through system prompts and custom code. Policies are checked at multiple workflow points and can be bundled with agents across frameworks. The SDK supports LangChain, OpenAI Agents SDK, Anthropic Agents SDK, AutoGen, CrewAI, and more.
Read the original article →