Summary

  • 2 minutes

You now have a responsible AI strategy for a Copilot Studio agent that connects Microsoft's principles to the decisions you make about moderation, models, guardrails, and disclosure on the standard and GitHub Copilot harnesses.

What you learned

In this module, you learned how to:

  • Turn the six responsible AI principles into design decisions, and separate the safeguards the platform provides from the ones you own.
  • Set content moderation on each harness, diagnose a filtered response, and place exceptions at the narrowest scope that needs them.
  • Judge a model's production readiness from its use tag, release type, safety notes, and the administrator controls that gate it.
  • Layer instructions, approved knowledge, tool boundaries, and approvals into guardrails.
  • Disclose the agent's use of AI and its limits to users.

Learn more