Anthropic bans ‘abusive or cruel behavior’ toward Claude in new usage policy

Anthropic has updated its Usage Policy for the first time in a year, and the line that grabbed attention is a new ban on “sustained and needless abusive or cruel behavior” toward Claude. The 2026 Usage Policy update takes effect on 12 November.

The abuse rule is narrow. Anthropic says it applies “only in extreme cases, where users repeatedly act cruelly toward our models, with no discernible purpose”, and not to “common versions of user frustration, pushback, dark creative themes, or model testing and research.” So swearing at a chatbot that keeps getting your spreadsheet wrong is unlikely to be the target.

How it will be enforced

The rule builds on a step Anthropic has already taken. Claude models can end rare conversations with persistently abusive users on Claude.ai and Claude Code, a feature that grew out of the company’s research into model welfare. Anthropic says that ability “will remain the primary enforcement mechanism.”

That leaves the question of account action open. The Verge reports that Anthropic did not say whether further measures, such as user bans, are planned. The Decoder points to the broader policy, which lets Anthropic warn users or throttle, restrict, suspend or terminate access for violations. Anthropic’s own post doesn’t tie the abuse rule to suspensions.

The company has also leaned into the question of whether its models have anything like an inner life. Kyle Fish, who leads its model welfare research, told The Verge in February that questions about consciousness and moral status are “serious ones that we’re investigating.” Rivals disagree: Microsoft AI’s code of conduct rejects the idea that models might deserve welfare or rights.

The other changes

Most of the update, according to Anthropic, clarifies existing rules rather than changing what it enforces. The main changes:

• Deceptive activity: rules once scattered across the elections, fraud, privacy and disinformation sections now sit in one section, “Do Not Engage in Deceptive Campaigns or Artificial Activity”. It covers political and commercial efforts to hide who is behind a message or to amplify content through fake accounts, plus building the tools for influence operations.
• Elections: the section, renamed “Do Not Undermine Democratic Processes”, now focuses on deceiving voters or disrupting elections. Anthropic has dropped its blanket ban on personalised vote and campaign targeting, citing legitimate uses such as nonprofits writing voter information in other languages or election officials sending ballot cure notices.
• Weapons: the ban now explicitly covers the software and components that make weapons work, including arming drones and other autonomous vehicles. Anthropic says it has seen people try to use its models to build guidance and control software for weapons.
• Surveillance and law enforcement: tracking people without their consent is prohibited, in real time or through analysis of previously collected data. Claude can’t be used to decide or recommend who to investigate, arrest or charge, or to build or improve surveillance tools. Consensual tracking such as fraud monitoring, content moderation, journalism and legal research stay permitted.
• High-risk uses: the human-in-the-loop and AI-disclosure requirements for health, legal, financial and similar uses are unchanged, but the section now lists which recommendations are covered.
• Physical hardware: when Claude is connected to equipment that takes autonomous physical actions and could cause injury, a qualified operator must be able to observe it and stop it, and the equipment must be able to hold a safe state if Claude disconnects.
• Supported regions: Anthropic clarified how it enforces its policy, which bars use by people physically located in unsupported regions, entities incorporated or headquartered there, and entities majority-owned or controlled from those regions.

The Verge notes that the surveillance and weapons rules may be modified for certain government customers if Anthropic judges the contractual safeguards adequate, and that the company has previously contracted with the US military.

Why it matters

For most users nothing changes in daily use. The abuse clause is mostly a statement of position: Anthropic is treating hostile treatment of a model as a policy matter, in the same document that covers weapons and election interference. The more practical changes are for developers and businesses building on Claude, especially in agent, robotics and high-risk consumer settings, where the clarified requirements decide what they can ship. As AI models take on longer autonomous work, as in Grok Bot’s move to Claude, those rules get more weight.