Announcing the addition on X, Mustafa Suleyman, CEO of Microsoft AI, explained that Portraits was developed in response to feedback from some Copilot users who expressed they’d be more at ease speaking to a face when using voice. This feature allows users to choose a portrait and pair it with a voice, aiming to make spoken interactions with the chatbot feel more authentic than written conversations.

This marks Microsoft’s newest effort to develop Copilot into a more engaging character to interact with, following the announcement of a similar feature known as “Copilot Appearances” in July. The key distinction is that Portraits is based on human likeness rather than an animated cartoon blob.

“This experiment uses VASA-1, an advanced AI technology created by Microsoft Research to facilitate real-time, visual AI conversations,” Microsoft shared on the Copilot Discord server. “This innovation enables us to produce natural facial expressions, head movements, and lip-sync from just a single image — bypassing the need for complex 3D modeling.”

Microsoft stated on the Copilot blog, “Each portrait is purposefully stylized rather than photorealistic, providing visual variety while maintaining the same trusted Copilot intelligence and safety features you depend on.” Access is limited to a “select set of users” aged 18 or older, with daily and session-based time limitations imposed alongside “clear indicators of AI technology interaction.”