Microsoft 365 Copilot licensees can now “adjust harmful content protection”, a configuration option controlled by IT Administrators & designed to support roles such as law enforcement, legal personnel, social workers, and medical staff by enabling them to use AI effectively in real-world scenarios. In some cases, these roles & fields may require reviewing or generating sensitive or potentially inappropriate language. (Ex; Medical reports, criminal evidence, or social work case documentation)
To support these job requirements, the feature allows for the ability to adjust harmful content protection settings so that Copilot Chat can respond to queries about harmful content when it’s fit for purpose. Access is restricted by administrators to approved personas only, ensuring it is used strictly for legitimate work purposes.
Adjusting harmful content protection does so exclusively for Copilot Chat & text responses & it will not disable “core responsible AI protections” including prompt injection defenses, copyright safeguards, biosecurity & image protection. Policy changes also do not affect image generation or agents.
When granted the ability to adjust harmful content protection, users can use a toggle to enable or disable harmful content protection in Copilot Chat. The menu includes a Harmful content protection setting, as shown in the following screenshot:
The administrative control is available to Microsoft 365 Copilot licensees & is available for both Commercial as well as Government/GCC Microsoft 365 Copilot cloud instances. For more details, read:













You must be logged in to post a comment.