The AI video news that matters to creators, in two minutes. Every item checked against a dated source before it goes up.
Each Friday we pick the handful of AI video developments that change what you can make, or how you should publish it, and say why in one line. Older editions stay below as an archive.
Kuaishou's new model is rolling out with longer single-pass shots, more keyframes and speech in nine languages.
More expressive voiceover in 90+ languages, and a voice clone now needs about 10 seconds of audio.
Editors can generate, restyle and upscale footage without leaving their timeline.
New Gemini-powered tools can trim dialogue, sync clips and reorder footage from a plain request.
Likeness detection will soon flag AI imitations of a creator's voice, not just their face.
A new law requires a clear disclosure on video or audio ads that use AI-generated performers.
New bills revise the provenance rules first set by SB 942. Worth a look if you publish AI video commercially.
Roll any shot as a quick 480p draft, keep the take you like, then finalize that exact take at 1080p.
Runway says its video tools now work inside OpenAI's Dots agents, which can plan shots on their own.
The AI video spinoff is preparing a Hong Kong listing, one more sign this is now a mainstream industry.