Microsoft patents AI that rewrites live speech and lip sync in real time

On November 2, 2026, Futurism reported that Microsoft has filed a patent for an AI system capable of transforming live video in real time, altering both a speaker's words and facial appearance. The filing was first spotted by patent-tracking platform Patentlyze. In the patent, the inventors describe the technology as a "near real-time artificial intelligence (AI) enhancement of video or augmented reality (AR) presentations of live events."

According to the patent text, the system relies on AI agents to identify, delete, or replace specific words spoken by a speaker, while simultaneously editing the speaker's face and mouth so that the visuals match the modified speech. These agents operate according to preset parameters or user profiles, and the system applies smooth transitions to the video before and after each alteration, so the edit is not visibly jarring on screen. The patent also mentions potential applications such as real-time translation.

The patent filing has been disclosed so far, with no confirmation of whether it has been granted or of any concrete product plans. Futurism's original snippet of the article states that there is simply no way such a system would not be abused almost immediately, though this criticism is Futurism's own editorial judgment.