A GitHub project called Reverse Engineer Anything went viral as people used AI to tear apart the proprietary software on smart TVs, baby monitors and dead hardware, rebuilding it without the ads and locks.
Since Saturday, everyone seems to have been doing the same thing: replacing the operating system on their smart TV. Not with code they wrote, but by having AI take the manufacturer's software apart and write a version without the ads and bloat.
Features the maker locked away, prised open
People turned abandoned smart picture frames into AI calendars, stripped every ad out of baby monitors, and brought bricked Lululemon mirrors back to life. Dozens of TVs got AI-written software that simplifies the interface.
One user, Gathin, posted that he fed his parents' TV model number to Claude Code. The AI found the TV on the wifi, removed 25 Xiaomi packages and wrote a new launcher. The post was reportedly viewed more than six million times.
One project turned it into an assembly line
What lit the fuse was a GitHub project called Reverse Engineer Anything, which pulled in more than 342,000 stars in a matter of days. Its pitch is blunt: see a feature in another app you want? Ask your agent to inspect it. It can explain how the feature works without the source code and build a version for you.
It is not confined to desktop software. A self-described guitar pedal maker claimed on Hacker News that he had 'one-shot decompiled' three competitors' products in 72 hours and could in theory box and resell them. Anyone with a Claude subscription, he said, can now rip off any competitor's best algorithm.
The mess after the fun
Mike Nolan, who studies the politics of open source, cautions that while AI has lowered the barrier to recreating functionality, software still needs maintaining. These clones are likely unreliable, and few companies will dare adopt them.
Worse, much of the AI output is not the original code at all. It is full of bugs the person using the AI cannot read or fix, and it gets installed on devices and dumped into communities anyway, faster than anyone can vet it.
Nolan also says engineers are already saying software-as-a-service is dying as a business model.
In the short term, expect half-baked hardware startups shipping cloned software, restaurant ordering screens running pirated systems with bizarre bugs, and internet-connected devices holding code nobody understands. Small companies getting undercut by rivals who cloned their product with AI is not a far-fetched outcome.
Why it matters
Once the barrier is a Claude subscription, the line between reverse-engineering and theft all but disappears. The losers will not be the big platforms but the small teams whose entire product was one algorithm or one app, now easy to lift.
Curated from high-quality sources, with concise summaries and key takeaways.
