AI
Anthropic writes model-abuse limits into its usage policy
Anthropic updated its usage policy on 8 October to bar “sustained and needless abusive or cruel behavior” toward its models, formalising a step it had already taken: allowing Claude to end a rare subset of conversations with persistently abusive users, a power the company says remains its main enforcement mechanism. Anthropic said the rule targets repeated mistreatment without a discernible purpose and excludes ordinary frustration, criticism, creative work and safety testing. The update takes effect on 12 November, alongside new limits on surveillance and propaganda uses.