
Always-on AI television where the audience votes on the next scene.
The live edge of generative media
Find live AI-generated channels, audience-directed shows, interactive worlds, and the models powering them — with verified status, last-checked times and sources.
Browse platforms, models, categories…Only streams confirmed by a public detector within the last hour appear here. Status snapshot: .
Editorial selection, not a popularity ranking. Trending appears only once the site has enough traffic to compute it honestly.

Always-on AI television where the audience votes on the next scene.

A single endless AI channel where the chat queue decides the next fifteen seconds.

A Kick livestream that can show red LIVE from Kick’s API — LIVE here means the channel is broadcasting, not that the feed is verified AI-generated.
The brand focuses on AI-native live media; the data layer also covers interactive worlds and the infrastructure underneath them.
Channels whose picture, narrative or characters are generated by AI while the broadcast is running, rather than pre-rendered and looped.
Continuous AI programmes where viewers steer what happens next through prompts, votes, choices or chat.
Always-on channels that generate the next segment of video while the current one plays, with or without audience input.
Experiences where a model renders an explorable world frame by frame in response to player input.
Systems that restyle or regenerate a live camera feed or an existing video in real time.
Model hosts, inference platforms and APIs fast enough to power live or interactive generative media products.
World model (research preview)
Video generation (post-trained variant)
Multimodal generation (video, image, audio)
Event dates are when things happened; they are kept separate from when we recorded them.
Wildlife World, Artificial TV, Directed By Public, H3 Story Lab, Renderdown, ALIBI and APOB Live were added as published directory pages with conservative LIVE rules.
Charlie Clark published Diiverge, a shared browser world where clicking a hotspot forks the scene and carved paths stay for later visitors. Walking existing ground is free; carving new scenes is paid.
World Labs published Atlas on 1 September 2026 as an omni world model for spatial intelligence, opening early access with select partners rather than a public live deployment.
Pieter Levels published Infinite Slop, a continuous AI video channel that generates the next clip from the top-voted chat prompt. A dedicated domain followed the next day.
fal announced H3 Max, a post-trained variant of MiniMax H3 that generates a five-second clip in about three seconds. It is the model the live channels we currently track are built on.
Entries ordered by the date they passed verification here, and the explainers that put them in context.
An AI-generated fan parody of SpongeBob SquarePants. The channel rates the stream for ages 16 and over (explicit and mature content).
Viewers chat with AI characters on Twitch and can create their own characters. The model is not publicly disclosed and is unverified.
A Japanese-language AI VTuber on Twitch who reads the chat and answers viewers directly.
This is an AI-generated satirical debate; the voices and likenesses are synthetic, and it is not Donald Trump or Joe Biden themselves. Chat viewers propose the topics.
An AI VTuber on YouTube whose stream title invites viewers to talk via chat.
A plain-language definition of AI live streams, how continuous generation works, the formats that exist today, and how to tell a real one from a loop.
The signals that prove an AI channel is on air, ranked by trust, and the rules that stop a directory from showing a false LIVE label.
A reachable website is not a live broadcast. Red LIVE requires a public stream or status endpoint confirming output within the last hour; otherwise we show Unverified or Stale. Read the methodology.
Each fact field traces to a public source with an observation timestamp. Where a fact cannot be confirmed we write “Not publicly disclosed” rather than guessing.