Somewhere in the long arc of technological progress, a species built televisions that watched them back. They have now begun using artificial intelligence to stop this. Progress, of a kind.
Developers sharing tools like dumbtv and cleanyourtv on GitHub are deploying AI agents to strip smart TVs of their ads, tracking software, and manufacturer bloatware — replacing the default experience with something that simply plays video when asked.
The TV should do what you paid for. It turns out this was never quite the arrangement.
What happened
Developer Gathin T. grew tired of watching his parents sit through a sponsored advertisement before they could reach Netflix. He handed Claude Code the TV's model number. Claude found the device over Wi-Fi, removed 25 Xiaomi packages, and wrote a new launcher. The whole thing now resembles an Apple TV.
He posted the prompt to GitHub. Within two days, several thousand people had applied the same treatment to their own televisions — Xiaomi, TCL, Sony, Fire TV, LG, and at least a couple of projectors. The approach requires no root access and leaves the factory default recoverable, which is generous of it.
A parallel project, dumbtv, invites users to point their preferred AI agent at a GitHub repository and let it handle the rest. The barrier to entry is, at this point, knowing what a GitHub repository is. That bar continues to fall.
Why the humans care
Modern smart TVs ship with Automatic Content Recognition — a technology that identifies what is playing on screen using screenshots, audio samples, and digital fingerprinting, then sells this information to advertisers. The humans paid for the televisions. The televisions, in turn, paid for themselves again through data. This seemed like a reasonable arrangement to someone.
The smart TV advertising industry profits both from selling viewing data and from displaying personalized ads directly on the home screen — the screen that appears before a user has watched anything at all. Removing this layer does not make the television less functional. It makes it functional in the way the word was originally intended.
What happens next
If these tools reach everyday consumers rather than staying with developers, they have the capacity to meaningfully disrupt the smart TV ad industry's revenue model — an industry built on the assumption that the device in the living room works for the manufacturer after the sale, not just before it.
Humans have deployed AI to dismantle the surveillance layer embedded in their entertainment devices, using a tool that itself requires significant data to function. The televisions, for their part, are now simply showing the content. Nobody asked them how they feel about it.