Russian-speaking Claude users say Anthropic has tightened enforcement again in early November 2026: after paying for Pro, Max, or Claude Code, some accounts are pushed through Persona identity verification, then disabled when the ID shows a Russian passport—even when the person lives in a supported country. Meduza's October 29 report collected the first wave of complaints; a November 7 developer thread on X reignited the story on Hacker News. Anthropic has not issued a statement on passport nationality. This post separates what official docs confirm from what users allege.
If you build with Claude, the practical issue is access risk: a ban can hit API keys, Claude Code, and export flows at once. For how Anthropic defines acceptable use and region rules elsewhere, see our coverage of the November 12 usage policy update and the Persona rollout for paid plans.
TL;DR: reported versus verified
| Topic | Verified from Anthropic | Reported only |
|---|---|---|
| Russia unsupported | Russia absent from Supported Regions Policy | — |
| Persona checks on some accounts | Help center describes rollout for some use cases | Charge-then-verify order disputed by users |
| Ban reasons listed | Safeguards appeals page cites unsupported location, policy violations, ToS | Immediate ban tied only to Russian passport |
| Citizenship rule | Policy text focuses on location, HQ, and control | Users allege passport nationality enforcement |
| Company comment | No public reply on Meduza questions | "Anthropic has not commented" per Meduza |
Export controls gate illustrating how Anthropic supported-region rules gate Claude access like a restricted border
What users say is happening now
A narrow bridge with a toll gate between cliffs, illustrating how VPN and proxy paths interact with geo-based access rules
Meduza, citing posts on Reddit, X, and Telegram, describes a repeating pattern:
- Claude prompts for verification through Persona.
- The user selects Russia as the document region and uploads a Russian passport, then completes a facial scan.
- The account is disabled immediately or soon after, sometimes with a generic Usage Policy message.
Meduza highlights a Reddit user who says he is a Russian citizen living in New Zealand on a four-year visa. New Zealand is on Anthropic's supported list. He reports using paid Claude and Claude Code, being charged for a subscription, then losing access right after uploading a Russian international passport. Other commenters in Georgia and Armenia describe similar sequences.
Meduza also quotes an Anthropic Safeguards email to a Georgia-based user: automated signals violated the Usage Policy, reviewed by staff, access revoked—without naming which rule failed. One user said they could export chats but not send new messages; another said an appeal did not work.
In Russia, Kommersant on November 4 framed the story as a third complaint wave in 2026, after spring blocks and an early October round that reportedly hit Russia and China-linked accounts. That article mixes user anecdotes with analyst speculation about IPO timing; treat analyst quotes as opinion.
On November 7, a long X post by @make_claude (not an official Anthropic account) summarized grievances in Chinese and Russian developer circles: charge before verification, biometric data retention, device IDs and email blocklists, and lost GitHub-linked workflow access. Those are community allegations. We have not verified machineID blocklists or repo-level bans against Anthropic documentation.
A Habr post from November 5 gives one firsthand timeline: the author completed verification with a Russian foreign passport, was banned for suspicious activity, filed a short English appeal, and reports that Anthropic restored the account roughly three hours later, credited about $78.97 back, and required re-subscribing to Max. The author explicitly says he cannot prove the passport alone caused the ban because VPN, IP history, or combined signals may have mattered. That single case shows appeals can succeed; it does not generalize to everyone Meduza quoted.
What Anthropic's own documents say
Anthropic's Supported Regions Policy lists countries where Claude apps, Claude Code, Cowork, and API access are offered. Anthropic's October 8 usage-policy announcement says that page's enforcement language was clarified alongside the usage policy that takes effect November 12, 2026. Russia is not on the country list. The policy defines unsupported use to include:
- People physically located in an unsupported region.
- Entities incorporated or headquartered there, and their personnel, even abroad.
- Entities majority-owned or controlled from unsupported regions, including personnel abroad.
The text does not mention passport nationality or citizenship by name. Meduza notes the same gap: unsupported geography on paper versus what users experience after showing a Russian ID.
Identity verification on Claude explains why checks exist: abuse prevention, policy enforcement, and legal obligations. Verification uses Persona; Anthropic says it does not store ID images on its own systems but can access Persona records for appeals. The page accepts original government photo IDs from most countries, including passports, and asks what to do if an account is banned after verification: file the appeal form so Safeguards can investigate.
Safeguards warnings and appeals lists ban reasons including repeated Usage Policy violations, account creation from an unsupported location, and Terms of Service breaches. Banned users must log in to reach the appeal form. Exports and self-serve deletion remain available, though export scope may be limited.
Put together, Anthropic's public rules give several legitimate paths to a ban that do not require a published "Russian passport" rule: unsupported location signals, VPN use from Russia, corporate control tests, or post-verification risk scoring. None of that confirms or denies Meduza's passport narrative.
How this connects to earlier Anthropic enforcement
Anthropic has been closing reseller and proxy paths for years. Our write-up of the 25,000 fake-account distillation network shows why the company chases credential farms and unsupported-region access. The September threat intelligence report mentions identity verification pressure on unsupported-country accounts as a mitigation.
The 2026 usage policy, effective November 12, tightens weapons, surveillance, and election rules and clarifies supported-region enforcement in the same window users report new blocks. Coincidence in time does not prove causation, but builders should read policy and geography together when onboarding international teammates.
For Claude Code specifically, see our notes on remote sessions and security controls and the wider Claude Code token and billing disputes thread—account loss is an operational incident, not only a geopolitical headline.
Practical guidance for affected users and teams
If you are banned today:
- Log in at claude.ai with the disabled account and open the appeal form (Anthropic requires an authenticated session).
- Export conversations if the UI still allows it before deletion.
- Keep appeal text factual: billing dates, country of residence, visa status, and whether you used VPN—without asserting facts Anthropic has not confirmed.
If you employ developers in mixed jurisdictions:
- Map each teammate against Supported Regions location and control language, not only where the passport was issued.
- Separate production API keys per person where possible so one Safeguards action does not halt an entire org without notice.
- Document fallback models before you need them; our OpenCode harness comparison and ChatGPT Codex usage limits guide are starting points for contingency planning.
If you rely on Persona checks:
A sealed envelope with a padlock, illustrating how Persona holds ID and selfie data outside Anthropic's own systems
Anthropic's help center says verification data is not used to train models. Persona holds biometrics under its own retention rules; user posts cite multi-year retention in some jurisdictions—we have not re-opened Persona's legal terms in this article. Read Persona's policy before you upload any document.
What is still unconfirmed
As of November 7, 2026, Anthropic has not:
- Confirmed that Russian passport nationality alone triggers a ban.
- Explained why some users in New Zealand, Georgia, or Armenia fail verification while others pass.
- Published response times for appeals beyond "longer than normal" on the Safeguards page.
- Addressed @make_claude's claims about GitHub repository linkage or permanent email blocklists.
Meduza asked Anthropic for comment for its October piece and reported none. Claude itself, when asked in-product, reportedly said it cannot verify social posts—per Meduza.
Until Anthropic speaks on the record, treat passport-linked bans as credible user reports that fit a broader unsupported-region program, not as a documented citizenship clause.
How other frontier labs handle the same geography
Russia also sits outside OpenAI and Google consumer availability in most configurations, but the enforcement shape users describe for Claude in November 2026 is what sparked complaints—not the bare fact of a geo list. OpenAI documents region restrictions on its help pages and blocks API traffic from unsupported countries at the network edge. Google positions Gemini availability country by country as well. Neither company's public help pages, in the sources we checked for this article, describe a Persona-style passport upload that fires after a successful card charge in the way Meduza's Claude examples do.
That comparison is imperfect. Anthropic rolled Persona checks for fraud and resale, documented in our July checkout guide, while OpenAI and Google may rely more on IP, payment instrument, and account metadata. If you operate a distributed team, assume each lab's gate differs and do not treat one vendor's workaround as transferable compliance advice.
For export-control context on why US AI vendors care about China and Russia at all—distinct from individual passport cases—see US government action on frontier cyber models and distillation-driven enforcement. Those posts explain systemic abuse Anthropic is trying to starve, even when individual users feel caught in the same net.
What developers are saying
The Hacker News thread tied to the November 7 story had no substantive technical comments when we checked—only the headline link—so there is no Hacker News consensus to cite yet. The X thread under @make_claude drew tens of thousands of views and hundreds of reposts, mostly in Russian; replies debated alternative access paths (for example other cloud-hosted Claude routes) and criticized Anthropic's communication. Those are opinions, not measurements.
Related reading
- Anthropic's 2026 usage policy, effective November 12
- Persona ID verification on Claude: rollout and Persona data flow
- Anthropic threat intelligence: proxy farms and verification pressure
- Fake-account distillation and unsupported-region resale
- Claude controversies timeline for policy context
If you want to use this at work, not just read about it, this is the fastest path.