Use AI responsibly with Dynamics 365

  • 6 minutes

Using AI features effectively means understanding not just what they can do, but how they're built to protect you, your customers, and your data. Microsoft designs AI capabilities in Dynamics 365 with responsibility at the core—not as an afterthought.

Microsoft's responsible AI principles

Microsoft's approach to responsible AI is guided by six core principles: fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability. Before any Copilot feature or AI agent ships, it passes a responsible AI review and red teaming—where teams actively try to find ways the system could fail or be manipulated. For more, see Microsoft Responsible AI.

How your data is protected

When using AI features in Dynamics 365, your data stays yours. Microsoft doesn't share it with third parties or use it to train models without your consent. Copilot respects your existing permissions—it only surfaces information the current user can already access. All data is encrypted and stays within the Microsoft Cloud trust boundary. Microsoft monitors for abuse through automated processing but doesn't manually review your conversations.

Note

Some AI features may process data outside your primary region to reach Azure OpenAI Service. This is why the Move data across regions prerequisite exists in the AI Hub. You're always asked to explicitly accept this before AI features are activated.

AI outputs require human judgment

Generative AI is powerful, but it's not infallible. Copilot and AI agents produce suggestions, drafts, and summaries, not guaranteed facts. Before you send a Copilot-drafted email, share an AI-generated case summary with a customer, or act on an agent's recommendation, take a moment to review the output.

A few things to keep in mind:

  • AI responses aren't always accurate. Copilot does its best to ground responses in your actual data, but it can still produce content that's incomplete, imprecise, or out of date. Always verify before acting.
  • Cited sources help. Where applicable, Copilot cites the sources behind its responses so you can check them yourself.
  • You're in control. AI features in Dynamics 365 are designed to assist, not replace, human decision-making. The agent researches the lead—you decide whether to engage. Copilot drafts the email—you decide whether to send it.

Tip

Treat AI output the same way you'd treat a first draft from a new colleague: useful as a starting point, but worth a second look before it goes to a customer.

Responsible AI resources for Dynamics 365

Each Dynamics 365 customer engagement app has its own Responsible AI FAQ that covers how specific features work, what data they use, and what their limitations are. These are worth bookmarking for your team, especially when rolling out new AI features to end users.