Zoom has announced the launch of photorealistic AI avatars, which will appear on the service in December 2025. They will allow users to participate in video calls even if they are not ready to turn on the camera.
To create an avatar, simply take a picture in the application or upload a photo. After generating the image, you can "dress up" in various professional costumes. During a call, the system will track the owner's movements and facial expressions, synchronizing them with the virtual display in real time.
Zoom promises safeguards. The Live Camera Authentication feature will check if the face in the camera matches the uploaded photo, and a flag will appear in the participant’s window that the AI is being used. According to Chief Product Officer Smita Hashim, these processes may change before release.
The update will also bring a live translation feature: the algorithm will broadcast the speaker's speech in nine languages, including English, German, Chinese, French, Spanish, Arabic, Japanese, Portuguese, and Italian, so that participants can hear each other in their native language without delay.
At the same time, the company is expanding the capabilities of AI Companion: the assistant will be able to join offline meetings and sessions in Microsoft Teams or Google Meet, take notes and create tasks, as well as take on the task of scheduling meetings and creating clips.
Thus, Zoom is consistently approaching the vision of CEO Eric Yuan, who a year ago spoke about a future where "digital clones" will perform some routine tasks instead of people.