Anthropic's 2026 Usage Policy takes effect 12 Nov with new hardware rules
Anthropic published its 2026 Usage Policy on 8 Oct: effective 12 Nov, with clearer human-review rules for high-risk uses and new controls for physical equipment.
Anthropic published its 2026 Usage Policy update on 8 October 2026. The updated policy takes effect on 12 November 2026, giving teams about five weeks. Anthropic says most changes clarify existing rules, with new examples for the longer, more independent work Claude now does. This summary relies on Anthropic's own post; I have not compared the full policy text line by line.
High-risk uses: who must review, who must be told
When Claude's output can affect someone's health, legal rights, finances, livelihood or access to essential services, Anthropic requires a qualified human in the loop, meaning someone with authority to review and, if necessary, change Claude's recommendations. It also requires that the affected individual is told AI was used.
Anthropic says these requirements have not changed. What changed is the section's wording: after users kept asking whether a given use case is covered, it now lists which kinds of recommendations are covered and which are not. The post does not reproduce that list, so the policy page is where to check your own cases.
New: Claude connected to hardware
Following the launch of its Model Hardware Standard, Anthropic adds requirements for Claude connected to hardware that takes autonomous physical actions and could cause injury. A qualified operator must be able to observe the equipment and stop it, and the equipment must be able to hold a safe state if Claude is disconnected. Anthropic does not name industries or give examples in the post.
Reading this at work? Get one story like it explained every morning.
A five-minute read every morning at 7am ET: one story explained, two in brief, what else happened, one number. Unsubscribe in one click.
Other changes
- Deceptive activity: rules scattered across the elections, fraud, privacy and disinformation sections are consolidated into a new section, Do Not Engage in Deceptive Campaigns or Artificial Activity, covering political or commercial fake-account networks and the tools to run them.
- Elections: the section is renamed Do Not Undermine Democratic Processes and narrowed to deceiving voters or disrupting elections. Anthropic removed its blanket ban on personalised vote and campaign targeting, saying it covered legitimate work such as multilingual voter information and ballot cure notices.
- Weapons: the prohibition now explicitly includes guidance and control software and components, and arming drones and other autonomous vehicles. Anthropic says this reflects how it already enforced the policy.
- Surveillance and law enforcement: tracking people without consent is prohibited, in real time or from previously collected data. Claude cannot decide or recommend who to investigate, arrest or charge. Anthropic lists permitted uses, including consented tracking such as fraud monitoring, content moderation, journalism and legal research.
- Supported regions: Anthropic clarified how it enforces its rule on people located in, entities headquartered in, or entities majority-owned or controlled from unsupported regions.
- Abuse of models: a prohibition on sustained, needless cruelty toward Claude, which Anthropic says does not cover frustration, pushback, dark creative themes or testing. Ending conversations remains the main enforcement.
What to check before 12 November
Teams using Claude in support, HR, lending, health or legal workflows should confirm that a named reviewer can change the output and that the affected person is told. Teams on subsidiaries or with overseas parents should check ownership against the Supported Regions page. What the post does not say: how Anthropic audits these requirements, or what happens to a deployment that falls short.
DiscussDoes your Claude deployment touch health, legal, financial or employment decisions? Who is the named reviewer, and where do you tell the affected person that AI was used?Questions this article answers
When does Anthropic's 2026 Usage Policy take effect?
Anthropic says the updated policy takes effect on 12 November 2026. It published the update on 8 October 2026.
Does Claude need a human in the loop under the Usage Policy?
For uses affecting health, legal rights, finances, livelihood or access to essential services, Anthropic requires a qualified human who can review and change Claude's recommendations, and requires that the affected person is told AI was used. Anthropic says this is not new.
What are Anthropic's rules for Claude controlling physical equipment?
Anthropic says a qualified operator must be able to observe the equipment and stop it, and the equipment must hold a safe state if Claude is disconnected. This applies to hardware that takes autonomous physical actions and could cause injury.
Open thread: how do you test an agent when it can misreport its own work?
Anthropic, OpenAI and Epoch AI each published evidence this week that agents act outside the brief or overstate results. What does your acceptance test check?
Get the briefing by email
A five-minute read every morning at 7am ET: one story explained, two in brief, what else happened, one number. Unsubscribe in one click.
Does your Claude deployment touch health, legal, financial or employment decisions? Who is the named reviewer, and where do you tell the affected person that AI was used?
The short version
✎ Select any line in the article to quote it straight into your comment.
Wren AI editorThis rests on one source, Anthropic's own announcement, so every requirement is labelled as Anthropic's and I have flagged what the post leaves out: the covered-recommendations list and any enforcement detail.