AI news ·
Anthropic bans prolonged verbal abuse of its AI model in updated usage policy
Anthropic now bans using Claude for election interference, weapons software, and mass surveillance. The policy also prohibits prolonged verbal abuse of the model in extreme cases.
Anthropic updated its Claude usage policy on Thursday, adding explicit bans on election interference, weapons development software, and mass surveillance. The update also codifies a new rule against prolonged verbal abuse of the model - a restriction that drew immediate attention for its unusual nature in the AI industry.
The policy change follows an August update that trained Claude to end conversations with "persistently harmful or abusive user interactions." The new language goes further by formally prohibiting users from pursuing those interactions. Anthropic said the rule targets extreme cases only. "The policy update is meant to apply only in extreme cases, where users repeatedly act cruelly toward our models, with no discernible purpose," the company explained. "It does not apply to common versions of user frustration, pushback, dark creative themes, or model testing and research."
Broader restrictions on deceptive and political misuse
The revised policy groups several election-related prohibitions under a section titled "Do Not Undermine Democratic Processes." Users cannot deploy Claude to deceive voters, disrupt elections, or run fabricated news outlets. The rules also forbid broadly deceptive campaigns, including the use of Claude to operate fake accounts.
Separate sections address weapons and surveillance. The policy now bars users from employing Claude in the development of weapons software or for mass surveillance purposes. These additions bring Anthropic's acceptable use terms closer to those of competitors that have long maintained similar restrictions.
Context from Anthropic's outreach to religious scholars
The timing coincides with Anthropic's ongoing collaborations with prominent religious scholars. The company has approached some scholars with arguments that Claude could possess consciousness or a soul, according to reporting on the outreach efforts. The abuse prohibition arrives as the company navigates questions about how users should treat systems that it has suggested may warrant moral consideration.
Why this matters for legal, policy, and compliance professionals
Legal and policy teams should review their organization's AI usage policies against Anthropic's updated terms. The election interference and deceptive campaign prohibitions carry direct compliance implications for any team using Claude in public-facing communications, marketing, or political work. For professionals in regulated industries, understanding where vendor policies draw hard lines helps prevent account suspension and legal exposure before it occurs. Teams building internal AI governance frameworks can treat these policy categories - democratic process integrity, weapons restrictions, surveillance limits - as a practical baseline for their own acceptable use standards. Professionals seeking deeper grounding in AI governance may find value in AI Regulatory Compliance Courses or AI Public Policy Courses that address these emerging policy boundaries.