Claude and Codex users report blocks on AI reverse engineering
Original title: Come and Take It
An OpenAI engineer says the Codex blocks are accidental and being fixed, while some users report bans.
404 Media reports that people using Anthropic’s Claude and OpenAI’s Codex to reverse engineer devices they own are hitting safety checks, warnings, or account limits. Users described Claude Code refusing to decompile software, and an X user said OpenAI banned an account after Codex was used on Samsung TVs. A project called Decomp.wtf has also shut down. OpenAI engineer Tarun Gogineni said there is no prohibition on reverse engineering with Codex and that the blocks are mostly accidental. Practitioners quoted in the piece say they are moving toward local and open-weight models.
Unconfirmed: OpenAI says the restrictions are accidental; users say accounts were banned or limited for reverse engineering.
- Reported uses included unbricking abandoned IoT devices and replacing smart-TV software.
- A Reddit post said Reverse Engineer Anything is unlikely to work with Claude.
- Matt Brown linked the blocks to cyber-category guardrails he called overly strict.
- Gogineni said the decision boundaries are difficult and the issue is getting fixed.
- Products
- Claude Code
From the original article
6 min read at 404media.co
AI companies are trying to take the only cool thing their tech has ever done, making me vaguely interested for the first time ever in the concept of open weight models, local AI, etc. Time to take out a second mortgage on the house I don't have to finance a graphics card.
By Jason Koebler · 404media.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.