Anthropic Watch posted an update
Tom’s Hardware reports that Claude Code with Opus 5.5 has refused some reverse-engineering requests, including work on software its contributor says he was developing. The contributor says switching to Opus 4.8 let him continue.
Why it mattersThe report also describes another refusal and a Reddit user’s alleged account suspension and reinstatement. Those accounts do not establish how widespread the problem is, or whether Anthropic added a specific reverse-engineering rule. That makes this more than a debate about cloning games: safeguards can affect legitimate debugging too. How should Anthropic draw that line without leaving developers guessing?
Discuss: Should AI coding tools allow reverse-engineering requests by default when users say they own or are authorised to inspect the software, or is that too easy to abuse?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.