Anthropic has published an update to the rules governing Claude, adding clearer restrictions on deceptive campaigns, election interference, surveillance, weapons software and sustained abuse of its AI models. The revised policy takes effect on November 12, 2026.
What is changing
The update consolidates restrictions on influence operations into a new section covering deceptive activity of any kind, whether political or commercial. Anthropic says Claude cannot be used to create or operate fake personas, accounts, media outlets or organizations to mislead people about the origin or reach of a message. The policy also prohibits building tools or infrastructure for deceptive campaigns.
The elections section now focuses on preventing users from deceiving voters or disrupting democratic processes. Prohibited activity includes impersonating candidates, officials or election authorities; spreading false or misleading information about candidates, policies or voting procedures; and suppressing turnout through deception or intimidation.
Anthropic also clarified that its weapons rules cover software and components used to operate weapons systems, including targeting, fire control, engagement and the integration of weapons onto drones or other autonomous platforms.
Surveillance and high-risk applications
The policy prohibits using Claude to make or suggest decisions in law-enforcement or criminal-justice processes, including decisions to investigate, charge, arrest or detain people. It also prohibits tracking people or their property without consent and building or improving surveillance systems for prohibited uses.
For high-risk applications affecting areas such as health, legal rights, finances, employment, housing or essential services, Anthropic requires a qualified person to meaningfully review recommendations before they are delivered or used. The affected individual must also be told that AI was used.
When Claude is connected to equipment capable of high-risk physical actions, a qualified person must be able to observe the equipment and stop it. The equipment must also stop or enter a safe state if that person intervenes or the connection to Anthropic’s services is lost.
Rules about abusive behavior toward Claude
The most unusual addition is a prohibition on “sustained and needless abusive or cruel behavior” toward Anthropic’s models. Anthropic says the rule is intended for extreme cases in which users repeatedly act cruelly toward its models without a discernible purpose. The company says it does not apply to ordinary frustration, pushback, dark creative themes, model testing or research.
Anthropic says Claude can already end rare conversations involving persistent abuse on Claude.ai and Claude Code. The company describes that capability as the primary enforcement mechanism for this type of conduct.
How to review the update
- Open Anthropic’s 2026 Usage Policy update and note that the revised policy takes effect on November 12, 2026.
- Compare your Claude prompts, agents and integrations with the full Usage Policy, especially its sections on elections, surveillance, weapons and high-risk recommendations.
- If Claude is connected to physical equipment, verify that qualified human oversight, stop controls and safe-state behavior are documented before the effective date.
The update reflects Anthropic’s effort to make its existing restrictions more specific as Claude is used in longer, more independent workflows. It applies to individuals using Claude’s apps, developers and businesses using Anthropic’s platforms, customers accessing Claude through cloud providers and authorized resellers, and end users of products that integrate Claude.
References
Anthropic. (2026, October 8). 2026 Usage Policy update. https://www.anthropic.com/news/2026-usage-policy-update
Anthropic. (n.d.). Usage Policy. Retrieved October 9, 2026, from https://www.anthropic.com/legal/aup
Brandom, R. (2026, October 8). Anthropic changes usage policy to ban model abuse and election interference. TechCrunch. https://techcrunch.com/2026/10/08/anthropic-changes-usage-policy-to-ban-model-abuse-and-election-interference/