Claude's New Usage Policy Bans Gaming Search and AI Answers

Anthropic published a new version of Claude's Usage Policy on October 8, 2026. It takes effect on November 12 and binds anyone who uses Claude through its apps, its API or a third-party product built on it. Among the banned uses, it now names networks of sites built to sway what search engines and AI assistants answer.

What we know

What it is: the yearly revision of the Usage Policy, according to Anthropic's post of October 8, 2026. The company says most of the changes clarify rules that already existed.
When: it takes effect on November 12, 2026. The previous version had been in force since September 15, 2025, per the archived copy.
Who it binds: people using Claude's apps or its API, customers reaching Claude through cloud providers and resellers, and the end users of products that integrate Claude, per the policy text.
A new section on deception: it gathers rules that were scattered and applies to political and commercial deception alike, the post says. It bans creating or running fake personas, accounts, media outlets or organizations. Its examples are sockpuppets, fake reviews and bots that claim to be human.
Search engines and AI: the same section bans manipulating the sources that search engines or AI systems draw answers from with content that misrepresents its origin, authorship or independence. The policy's example: networks of sites posing as unaffiliated sources that back up the same claims.
The reason given: Anthropic says it has seen state media outlets, government propaganda offices and commercial firms use Claude to run networks of fake accounts and fabricated news sites.
High-risk uses: the two existing requirements stay, the post says: a qualified person able to review and change the recommendation, and telling the affected person that AI was used. The section now lists eleven areas, among them legal, medical, finance, credit, insurance, housing and employment.
Chatbots: any consumer-facing chatbot or AI agent must disclose that it is AI and not a human, at least at the start of each session or in the product interface. Naming Anthropic or Claude is not required.
Agents: whoever builds or deploys an agent is responsible for what it does through tools, browsers or connected systems.
Treatment of the model: sustained and needless abusive or cruel behavior toward the models is now banned. The post says this does not cover ordinary frustration, dark creative themes or testing.
Other changes: the post also covers elections, weapons, surveillance, supported regions and equipment that takes physical actions without human approval.

What changes and what doesn't

Fake reviews were already banned: the 2025 version listed them among fraudulent practices. What changes is that commercial deception gets its own section and that the text names search engines and AI systems. The 2025 version did not mention them. That comparison of the two versions comes from this article, not from Anthropic.

There is another difference the post does not highlight. The 2025 version treated content generated and published automatically for an outside audience as a high-risk use case. The new list leaves it out. That is not a free pass: the policy still bans disinformation, spam and plagiarism.

The chatbot disclosure is not new either. The 2025 version required it at the start of each session; the new one also accepts a notice in the interface.

If Anthropic suspects a violation, it may warn the user or throttle, limit, suspend or terminate access, per the policy.

This is one AI provider's rulebook. It does not change any search engine's rules.

How to tell if it affects you, and what to do before November 12

Find out whether you are using Claude without knowing it. Your site's chatbot, a writing tool or a plugin may run on it. Ask the vendor which model sits behind it: the policy reaches end users.
Look at your chatbot. It has to say it is AI when the conversation starts or in the chat window itself. A label such as "AI assistant" is enough.
Don't manufacture reviews or testimonials. Asking AI for opinions that read like customers' is banned, and already was. Ask real customers for real reviews.
Revisit your plan for showing up in AI answers. Spinning up several "independent" sites that repeat the same claims is the example the policy gives. What is banned is misrepresenting origin, authorship or independence. Publish under your own name and brand.
Check whether you make recommendations about specific people. Screening résumés, approving a loan or assessing a symptom requires a qualified person to review first, and the person must be told. Explaining a law or a treatment in general terms falls outside, per the policy's exclusions.
Check your region. Access is limited to the countries on Anthropic's supported list, and use is also barred for companies controlled from an unsupported region.

This article is not legal advice: the policy is Anthropic's own document, and the law where you operate may require more.

Sources

Updates: if Anthropic corrects the policy text, moves the effective date or publishes enforcement guidance, it will be added here with a link.