Anthropic is updating its usage policy for Claude, with changes taking effect November 12. Multiple outlets report that the update adds explicit prohibitions on sustained, needless abusive or cruel treatment of the model, while indicating that normal criticism or everyday frustration is still permitted.
The policy also expands restrictions aimed at high-risk categories of misuse. Reported additions include tighter rules around election interference and deceptive campaigning, limits related to weapons software or weapons development, and prohibitions involving surveillance. One outlet also notes broader coverage of sensitive use cases, including health and financial contexts, as part of the reasons for the update.
Across the reporting, the common theme is that Anthropic is responding to evolving misuse patterns by clarifying what users must not request or do with Claude. The update is framed as the first significant set of policy changes in more than a year, with specific guidance that targets both abusive interaction and harmful, safety-critical requests.
