Microsoft has filed a patent for a technology that could fundamentally change how people communicate through live video. The proposed system, known as Microsoft AI video editing, would allow artificial intelligence to modify a person’s spoken words during a live video stream while simultaneously changing their mouth movements to match the edited audio.
The technology could remove unwanted words, correct pronunciation mistakes, eliminate offensive language and even replace expressions according to individual viewer preferences. Although the patent describes a system designed to make these changes almost instantly, it also raises important questions about authenticity, misinformation and trust in digital communication.
The patent application, titled Near Real-Time AI Enhancement of Video or AR Presentations of Live Events, was filed under US application number 20260341309.
How Microsoft’s Live Video AI Technology Would Work
According to the patent filing, the system would use artificial intelligence to modify both the audio and visual elements of a live presentation.
When someone speaks during a video call or live broadcast, the AI could identify words that need to be changed and replace them with alternative expressions. It would then digitally redraw the speaker’s mouth to synchronise the visible lip movements with the modified audio.
The process is designed to take approximately one to 10 seconds, allowing changes to appear while the video is still being streamed.
This approach goes beyond traditional audio editing tools, which generally require a recording to be processed before changes can be made. Microsoft’s proposed system would instead work during live communication, making the edited version appear more natural to viewers.
The technology could potentially be used in video conferences, online presentations, live broadcasts and augmented reality experiences.
What Words Could the AI Change?
The patent describes several possible uses for the technology, particularly improving the clarity and presentation of spoken communication.
Some of the proposed editing capabilities include:
- Removing filler words: The AI could eliminate expressions such as “um” and other unnecessary pauses.
- Correcting pronunciation: Mumbled, mispronounced or unclear words could be replaced with clearer alternatives.
- Filtering offensive language: Inappropriate or offensive expressions could be removed from a live presentation.
- Changing slang: Informal language could be replaced with more suitable wording depending on the audience.
- Personalising speech: Different versions of spoken content could be generated according to a viewer’s age, location or language preferences.
The patent also describes a separate capability involving live translation. Under this feature, the system could translate a speaker’s words into another language while adjusting their mouth movements to match the translated speech.
For example, someone speaking English could appear to speak another language during a live video presentation, with the AI synchronising the visible movements of their mouth.
Why the Technology Raises Misinformation Concerns
Despite its potential benefits, Microsoft AI video editing has raised concerns about how audiences could distinguish authentic speech from AI-modified content.
The most significant issue is that viewers might not realise that the words they are hearing were never actually spoken by the person on screen.
Unlike conventional video editing, where changes are usually made to recorded footage, the proposed system could modify speech as the video is being delivered. This could make it difficult for viewers to identify alterations, particularly if the technology produces convincing audio and lip movements.
The possibility of changing someone’s words also creates potential risks involving misinformation, impersonation and fraud.
For instance, an altered video could make a speaker appear to support a statement they never made. Similarly, malicious users could potentially exploit such technology to manipulate business discussions, public announcements or political communications.
These concerns are particularly important as AI-generated voices, realistic digital faces and synthetic videos become increasingly sophisticated.
Patent Includes Limited Protection for Original Recordings
Another detail attracting attention is the patent’s treatment of original recordings.
According to the reported filing, only one dependent claim specifically addresses retaining the original recording. This has raised questions about whether viewers or organisations would have access to an unmodified version of the conversation.
Preserving original audio and video could be important for verification, accountability and resolving disputes involving edited statements.
However, the patent’s claims alone do not establish how Microsoft would handle recording storage, disclosure or access if the technology were developed into a commercial product.
Is Microsoft Planning to Release the AI Technology?
It is important to understand that a patent application does not mean a technology is ready for public use.
Microsoft has not announced plans to release this particular system as a commercial product. The filing describes a proposed invention and its possible capabilities, but it does not confirm that the company will develop or launch it.
The application also reflects Microsoft’s wider investment in artificial intelligence, including developments across its Copilot ecosystem and healthcare-related AI research.
The Future of AI-Edited Live Communication
Microsoft’s proposed technology highlights how artificial intelligence could reshape digital communication. Automatically correcting speech, translating languages and improving presentation quality could make online conversations more accessible and professional.
The ability to change spoken words without viewers noticing introduces serious questions about consent, transparency and digital authenticity.
If such systems become widely available, clear safeguards may be needed to distinguish original speech from AI-modified content.
Microsoft AI video editing remains a patent concept rather than a confirmed product. Its development could offer useful communication tools, but it also demonstrates why transparency and trust will become increasingly important as artificial intelligence gains greater control over digital media.
