Anthropic has updated its Supported Regions Policy, which takes effect on November 12, 2026. The restrictions apply to Claude.ai, Claude Code, Cowork, the API, and Claude models accessed through cloud platforms and authorized resellers. The ban covers individuals in unsupported regions, companies with registration, headquarters, or control from such regions, and all their employees, regardless of their actual physical location. Russia, mainland China, Hong Kong, and Macau are not included in the list of supported regions.

image
image

What Happened

Anthropic has defined three categories of individuals prohibited from using Claude starting November 12, 2026. The first category consists of individuals located in an unsupported region. The second category includes companies registered or headquartered in such a region; the restriction applies to all their employees regardless of location. The third category covers companies directly or indirectly more than 50% owned or controlled by entities from unsupported regions. The update was announced in the official post "2026 Usage Policy update," published on October 8, 2026, and updated on October 14, with specialized news publications confirming the details on November 10.

Context

Claude's regional restrictions previously functioned as a geofilter, where access was determined by the user's physical location. The new policy shifts the verification to the corporate context—registration jurisdiction, headquarters, and ownership structure. Consequently, using a VPN or being in a supported country no longer guarantees access if the employer falls under the ban. The rule covers all access channels to the models: not only the web interface but also the API, cloud platforms, and authorized resellers. This means the issue concerns the availability of the models themselves as work and research tools, not just a single service. The vendor itself defines the geography, with the list of supported regions provided in its legal document.

Why This Matters for the Industry

For the industry, the structure of provider dependency risk is changing. While verification previously relied on geolocation, companies controlled access to models through point measures. Now, the criteria are jurisdiction and ownership structure, so compliance teams and developers using Claude in corporate products need to review who specifically has access to API keys and cloud instances, and verify counterparties with registration or capital from unsupported regions. For affected companies, launching new products on Claude is effectively closed, and existing dependencies require a backup plan. Discussions anticipate interest in multi-provider architectures and open-weight models; however, this is a predictive interpretation, not a consequence directly documented in Anthropic's documents.

Why This Matters for Users

For the individual reader, the outcome is formulated strictly: if you work for a company with registration, headquarters, or majority shares in an unsupported region, your Claude account may be blocked, even if you are personally located abroad in a fully supported country. The employer becomes the criterion, so location, VPN, and payment details do not change the situation. Before November 12, 2026, it makes sense to check whether your organization falls under one of the three prohibited categories and whether your work processes and personal projects are tied to Claude via the web interface, API, or accounts with authorized resellers.

What Is Still Unknown / Limitations

The main unknowns lie in the area of enforcement: available materials do not describe how Anthropic will verify indirect ownership or corporate control, nor whether such verification is built into the client onboarding procedure. The policy establishes a ban on use but does not guarantee the automatic blocking of every account, so the depth and timing of the actual application of the rules are not visible from the sources. Assessments regarding teams transitioning to open-weight models and the separation of infrastructure by independent model comparisons across jurisdictions are future interpretations, not established facts. The exact list of supported regions and current wording should be verified against Anthropic's official document on the date of publication.

Sources

Author

Look at AI, editorial team