Details
Meta's AI video research progressed through Make-A-Video (announced September 2022, research only), Emu Video (announced November 2023, research only, producing 4-second clips from text or image prompts), and Movie Gen (announced October 2024, a 30-billion-parameter model capable of generating up to 16-second HD video with synchronized audio). Movie Gen also demonstrated the ability to insert a person's likeness into generated video from a single photo, and remained in research and limited partner testing as of its announcement. Vibes, the consumer-facing feature, launched in the Meta AI app and on meta.ai in September 2025, initially in the US and expanding to Europe in November 2025; users can generate short videos from text prompts, remix existing clips, and cross-post to Instagram and Facebook Stories and Reels. For the early version of Vibes, Meta integrated models from Midjourney and Black Forest Labs while continuing to develop its own video models. Meta stated its models were trained on a combination of licensed and publicly available data, and applies an 'AI info' label to AI-generated content on its platforms. As of February 2026, Meta was testing a standalone Vibes app in Brazil and Mexico and announced plans to introduce a freemium subscription model for video creation.
Have evidence about Meta's AI practices? Submit a report.
Report a Sighting →