The correct completion is “cannot modify the settings of” because built-in agents such as the Researcher agent in Microsoft 365 Copilot are system-defined experiences. These agents are designed and managed by Microsoft to ensure consistent functionality, security, compliance, and alignment with responsible AI standards.
While users can interact with Researcher, provide prompts, and refine outputs during conversations, they cannot directly modify the core system settings or configuration of the built-in agent from Microsoft 365. Administrative configuration and deeper customization are available when creating custom Copilot agents through approved development tools, but default Microsoft-provided agents have controlled configurations.
Options such as modifying agent instructions, adding prompts, or deleting suggested prompts imply direct structural customization of the built-in agent interface, which is not supported in the standard Microsoft 365 Copilot experience.
This reflects a key generative AI governance principle in enterprise environments: system-level controls remain centrally managed to maintain data protection, compliance, and responsible AI implementation across the organization.
