#

Key takeaways

• Anthropic published a revised Usage Policy on October 8.

• The policy takes effect on November 12.

• Many changes are described as clarifications, not brand-new bans.

• The update covers deceptive campaigns, weapons software, surveillance, and high-risk decisions.

• It also adds rules for hardware-connected models and sustained abuse.

Anthropic publishes a revised Usage Policy

Anthropic says it published a revised Usage Policy on October 8. The policy takes effect on November 12. The company frames most changes as clarifications of existing restrictions.

The update comes as Claude undertakes longer and more independent tasks. Anthropic does not present the document as a new law or government rule. It is a company policy update from Anthropic News.

What changed in the policy

A new consolidated section covers deceptive campaigns and artificial activity. That includes fake accounts, concealed sponsorship, and influence infrastructure. Anthropic says these rules apply in both commercial and political contexts.

The company also clarifies prohibitions involving weapons software and components. It further specifies surveillance restrictions. One example is tracking people without consent.

Anthropic says Claude must not decide or recommend whom law enforcement should investigate, arrest, or charge. That is a clear limit on law-enforcement decision support.

High-risk decisions need human review

The update restates high-risk-use requirements in clearer terms. Anthropic says a qualified human reviewer is required when AI supports decisions about health, legal rights, finances, livelihood, or essential services. It also says an affected person must receive disclosure.

This section does not introduce a new category in the source material. Instead, it makes the existing requirement easier to read. The policy language is more explicit about review and disclosure.

Rules for physical systems and model abuse

Anthropic also adds conditions for models connected to hardware that can act physically. The operator must be able to observe and stop the equipment. The equipment must also hold a safe state if disconnected.

The policy addresses sustained, needless abuse of models as well. Anthropic says this provision targets extreme cases of repeated cruelty without a discernible purpose. It says ordinary frustration, pushback, dark creative work, and model testing are outside that scope.

Anthropic also clarifies its Supported Regions Policy. Anthropic says access remains barred for people physically in unsupported regions, entities based there, and entities majority-owned or controlled from those regions. The announcement does not establish any new regional availability.

What this update means

The main theme is tighter wording around existing limits. Anthropic says many provisions are clarifications of prior enforcement. That matters because it suggests the company is making expectations more explicit as Claude is used in more autonomous settings.

The source does not report independent verification of the policy effects. It also does not report customer reactions, enforcement outcomes, or quantified impact. Any reading beyond the published policy would be an assumption.

Morocco relevance

The source reports no Morocco-specific fact. For readers, the global lesson is that policy updates can tighten operational rules without changing the underlying product category.

Bottom line

Anthropic's revised Usage Policy focuses on clearer boundaries. It covers deceptive activity, surveillance, weapons-related software, high-risk decisions, and physical systems. The company presents the update as clarification first, and new restriction only where the text says so.