Full Breakdown
Anthropic Bans “Sustained and Needless” Abuse of Its AI Models
By Drooid · · How we work
Policy Update Details
Anthropic announced on Thursday that its usage policy now includes a prohibition on “sustained and needless” abusive behaviour toward its Claude models. The new rule adds “cruel” interactions to a list of forbidden conduct that already covers bullying, self-harm promotion and the creation of non-consensual intimate imagery. Anthropic says the ban will be triggered only in “extreme cases” of repeated abuse and will not apply to ordinary frustration, dark-themed creative work, or legitimate model testing. The update also expands restrictions on “deceptive campaigns,” explicitly barring the use of Claude to mislead voters or disrupt elections as U.S. midterm polls approach.
Official Statements & Responses
Anthropic’s spokesperson explained that the change follows the firm’s annual policy review and is intended to address emerging risks, noting that the company had previously blocked the use of its tools for weapon development. Microsoft AI director Mustafa Suleyman, who earlier criticised treating AI as human-like, responded on social media with a melting-face emoji, signalling disapproval without further comment.
Criticism & Opposition
Legal technologist Dr Barry Scannell warned that anthropomorphising AI could mislead the public:
“You can't be cruel to numbers and maths,” — Dr Barry Scannell, technology partner at Irish law firm William Fry,on LinkedIn
He argued that labeling numbers and mathematics as “cruel” creates a false sense of agency in the technology.
Data & Statistics & Broader Debate
Some commentators contend that politeness wastes computational resources, while others claim it can improve response quality and encourage more respectful human-computer interaction. The Collins Dictionary recently highlighted the resurgence of the pejorative term “clanker” for AI tools, reflecting a cultural pushback against overly familiar language.
What May Follow
Anthropic indicated that the policy will be enforced in “rare” cases, but the lack of a precise definition for “abusive or cruel” leaves room for interpretation. Industry observers expect further dialogue on whether AI etiquette should be codified, especially as generative models become more integrated into everyday workflows.