The thread links an X post saying a person who was reverse-engineering his own TV had been banned from OpenAI, and the poster adds that an open-source reverse-engineering tool called REA was blocked by Claude a day later, linking another X post that says Anthropic stopped Claude from doing it. Gadget Review reports that a user known as @log4j says OpenAI closed his account, labeling the activity "Cyber Abuse," after he used AI help to reverse-engineer a Samsung TV he owns, reportedly over concerns about its telemetry and ad data collection; he says the work was local and that he never used OpenAI to find vulnerabilities. The site notes his account hasn't been independently verified, that OpenAI hasn't explained the case, and that claims about Claude come from user posts, not any Anthropic announcement.

The top replies urge tinkerers to keep quiet until work is done, suggest moving to open models (one replies that some of those are pulling cyber training too), and argue that safeguards shouldn't be decided by corporations. About 3,500 upvotes in a 329,000-member community.