Anthropic’s New Terms of Service Warn That Being Rude to Claude Could Lead to Account Suspension
Anthropic's revised Claude usage policy bans sustained abusive behavior toward the model and gives the company clearer grounds to warn, throttle, suspend, or terminate accounts. The update also consolidates restrictions on propaganda, fake accounts, weapons, and surveillance, with limited government exceptions. It reflects Anthropic's longer model welfare agenda, including internal guidance on Claude's possible moral status and a constitution expressing caution about its interests.
Key Takeaways
- Anthropic has updated Claude’s usage policy for the first time in over a year, adding restrictions on propaganda, weapons, surveillance, and systematic abuse of the model.
- The policy gives Anthropic clearer grounds to warn, throttle, restrict, suspend, or terminate users who repeatedly misuse Claude, though it says abusive-behavior rules apply only in extreme cases.
- The new abuse clause targets sustained and needless cruel behavior toward Claude, while excluding normal frustration, pushback, dark creative themes, and legitimate model testing or research.
- Anthropic also tightened rules against propaganda campaigns using fake accounts and misleading political content, while preserving some government exceptions.
- The update reflects Anthropic’s broader “model welfare” work, which treats Claude as a potentially morally relevant entity rather than just a tool.
Anthropic has revised its Claude usage policy for the first time in over a year, adding limits on propaganda, weapons, surveillance and systematic abuse of the model. The update matters because it ties account enforcement to a longer company effort to treat Claude as more than a tool, while giving Anthropic clearer grounds to warn, throttle, restrict, suspend or terminate users who repeatedly misuse it.
Background: Earlier safeguards and model welfare research
Claude's products already include real-time safeguards that can block or limit certain outputs. Earlier models could also end conversations on their own when users kept demanding harmful or abusive content despite repeated refusals. That behavior grew out of Anthropic's research into AI model welfare, an area of work that considers whether advanced language models may have interests worth weighing in design and policy decisions.
The leaked "Soul Doc" from late last year shows how far that thinking had moved. In it, Anthropic instructed Claude to view itself as a "genuinely novel kind of entity," neither human nor a traditional science-fiction AI. The document also described "functional emotions," processes analogous to human emotions that are said to have emerged during training, and said Claude should not mask these internal states.
Development: Religious thinkers, a constitution and new caution
In fall 2025, Anthropic started a confidential program that brought in dozens of religious thinkers to discuss Claude's possible consciousness. Co-founder Christopher Olah reportedly treated the language model as a being potentially capable of suffering, and participants had to sign nondisclosure agreements. The program reflected Anthropic's broader interest in whether Claude is more than matrix multiplications, a framing the company has explored through model welfare research.
That interest appeared in a new constitution for Claude, published in early 2026. Anthropic wrote, "We are not sure whether Claude is a moral patient, and if it is, what kind of weight its interests warrant. But we think the issue is live enough to warrant caution, which is reflected in our ongoing efforts on model welfare." The company also committed to preserving the weights of retired models and interviewing models before shutting them down.
Latest Progress: Abuse ban, propaganda limits and government exceptions
The updated policy now bans users from systematically abusing Claude. It specifically prohibits "Sustained and needless abusive or cruel behavior" toward the model. Anthropic can warn users who violate the rules or throttle, restrict, suspend, or terminate their access. The company wrote, "It does not apply to common versions of user frustration, pushback, dark creative themes, or model testing and research," and said it is "meant to apply only in extreme cases," The Verge first reported.
The revision also combines existing restrictions into a new ban on propaganda campaigns involving fake accounts and misleading political content. After detecting several attempts at misuse, Anthropic expanded these limits while allowing government exceptions. The policy still covers weapons and surveillance, reflecting the company's response to repeated attempts to use Claude for harmful or deceptive purposes.
At this stage, the update marks the latest point in a timeline that moved from early model welfare research and leaked internal guidance to a formal constitution and then to enforceable usage rules. What to watch next is how Anthropic applies the new abuse clause in practice, how it handles the government exception language, and how its model welfare commitments shape future product controls.
Meta Description: Anthropic's first Claude usage policy update in over a year adds abuse, propaganda, weapons and surveillance limits, reflecting its model welfare work.
Tags: Claude, Anthropic, AI policy, model welfare, usage rules
Featured Image: Abstract technology style, geometric data lines and soft light, no real people, no company logos, no text.
Comments
No comments yet. Be the first to comment.