Microsoft unveils control standard for AI agents targeting developers

Microsoft unveiled an open-source standard for governing AI agents through policy files that define allowed and blocked actions, approval points, and logging scope.

The Agent Control Spec applies controls before input, around tool calls, and before final responses, creating a common governance layer instead of scattered prompts, code, and filters.

The spec is designed to work as a plugin-style SDK across multiple agent frameworks and tools, including LangChain, OpenAI Agent SDK, Anthropic Agent SDK, AutoGen, CrewAI, Semantic Kernel, Microsoft.Extensions.AI, and MCP.

For enterprises, the policy-based approach should make AI agent security, compliance, and auditing more consistent and easier to operate.