Blog

Identity Verification in Claude: Persona and Action Plans

Markdown

How Claude identity verification works, the documents and selfie required by Persona, the conditions for different plans, how to preserve projects, and how to choose an alternative client.

Read full article

Key takeaway

Claude verification is a physical document scan plus a live selfie through the Persona service; the process takes up to 5 minutes (according to Anthropic’s official help article, updated August 11, 2026). Free, Pro, and Max are subject to verification; Team, Enterprise, and API are not affected. Below are three action plans and an explanation of how to preserve access to your account.

What changed in Claude identity verification on July 8, 2026?

Updated October 2, 2026. Verification is no longer “coming into effect”—it is in effect. Anthropic’s identity verification help article was updated August 11, 2026: the wording is the same (“several use cases,” “when accessing certain features”), but this now describes an active process rather than a future announcement.

Key takeaway

On July 8, 2026, Anthropic’s updated Privacy Policy took effect, adding a “Verification data” category under which the company may require Claude consumer accounts to provide a passport and selfie. Verification data and verification requests are used “to verify your identity and comply with legal and safety obligations”—that is what the current version of the help article, as of August 11, 2026, says. Anthropic still does not publish the exact trigger for a specific account.

On April 15, 2026, Anthropic first published its identity verification help article—it then sounded like, “we may ask you to confirm your age and identity.” On July 8, 2026, the requirement was enshrined in the Privacy Policy. On August 11, 2026, the help article received its latest update as of October 2, 2026: the text about documents, Persona, and data protection remained unchanged, and a description of the process for organizations was added (a legal entity confirms its name, address, and registration number in addition to the document and selfie).

Effective July 8, 2026. In certain circumstances, we may ask you to confirm your age or identity.

— Anthropic Privacy Policy, https://www.anthropic.com/legal/privacy

As of October 2, 2026, Claude’s Support page describes the process the same way it did in August: which documents are accepted, how Persona processes data, and that Anthropic does not store IDs on its servers. The process has been technically ready and operational for three months. The only question is when Anthropic will show you a request specifically.

We are rolling out identity and business verification for several use cases, and you may see a prompt to verify your identity when accessing certain features, as part of our routine platform integrity checks or other safety and compliance measures.

— Claude Help Center, https://support.claude.com/en/articles/14328960-identity-verification-on-claude

Key dates to remember:

  • April 15, 2026 — Anthropic first published its identity verification help article.
  • July 8, 2026 — The new Privacy Policy took effect. From this point on, Anthropic has a legal basis to request a passport at any time.
  • August 11, 2026 — The latest update to the help article as of today: the procedure for organizations (business verification) was added, and the wording about “several use cases” was confirmed again.
  • “As part of routine platform integrity checks” — means verification may appear at any time without warning. The company has not disclosed the exact trigger, as of October 2, 2026.

If your account has already been asked to verify and you were denied, that is not the end of the road: an explanation of what to do next and how to appeal is collected in a separate guide, how to keep your Claude account.

Practical materials on models and agents: Ailoot documentation.

Who will verification affect, and who will it not affect?

Key takeaway

Verification applies only to Claude consumer plans—Free, Pro, and Max. Team, Enterprise, and API access through Platform remain unchanged; KYC is already built into the commercial terms there. This means that for Anthropic’s three plans, behavior differed on July 8.

Anthropic has divided users into two groups.

Group 1 — subject to verification:

  • Claude Free (free plan through claude.ai)
  • Claude Pro (personal subscription at $20 per month)
  • Claude Max (from $100 per month for heavy use, up to $200 on the 20x tier)
  • Claude Code on these subscriptions (no separate fee required; tied to Pro/Max)

Group 2 — NOT subject to verification:

  • Claude Team (from $20 per user when billed annually, for teams)
  • Claude Enterprise (large corporations, contract)
  • Claude API through Anthropic Platform (separate developer account)
  • Claude API through AWS Bedrock or Google Vertex (verification is handled through AWS/Google)

The logic is simple: where Anthropic has a contract with a legal entity, KYC has already been done. Where individuals work directly, there was no identity verification contract before July 8; now there will be.

This creates an interesting workaround, discussed in the section on Plan C. The option of using the API through your company formally falls under Commercial Terms, and no personal verification is required. It is a viable plan, but not for everyone.

How will the verification itself work?

Key takeaway

Anthropic will show you a “verify your identity” modal. The modal links to Persona. There, you scan your passport or another ID using your phone’s camera, take a live selfie in real time, wait a few seconds, and the result is returned to Claude. The entire process is said to take “up to 5 minutes.” Persona stores the ID; Anthropic does not.

The process is simple—three steps.

  1. You receive an invitation in Claude

    A window pops up on the home page or when you try to use a feature: “we need to verify your identity.” Without consent, you cannot proceed—the feature is blocked, or the entire account goes into pending status.

  2. The “Verify identity” button opens an embedded frame or the Persona external app. This is not Anthropic—it is a separate service Anthropic hired.

  3. You do two things

    Scan a document (you need a phone camera or webcam—a simple photo from an old screenshot is not accepted) and take a live selfie with the camera in real time. Persona compares the faces.

Persona processes data on your behalf under our instructions. Anthropic does not store ID images on its servers.

— Claude Help Center, https://support.claude.com/en/articles/14328960-identity-verification-on-claude

Documents accepted:

  • Passport (international or domestic, with a photo)
  • Driver’s license
  • State or provincial ID card (for the US/Canada)
  • National ID card (for EU countries, the UK, and others)

Documents not accepted:

  • Photocopies, screenshots, scans
  • “Photo of a photo” (for example, a screenshot of a passport on a phone screen that you photograph again)
  • Digital or mobile IDs (Apple Wallet, mDL)
  • Student, corporate, or non-government-issued IDs

Photocopies, screenshots, scans, or photos of a screen are not accepted. Digital or mobile identity documents are not accepted.

— Claude Help Center, https://support.claude.com/en/articles/14328960-identity-verification-on-claude

Age checks are handled through a separate service, Yoti, which looks at your face and estimates your approximate age. If you are under 18, the account is closed regardless of whether you have a passport.

Supported countries and documents

What Anthropic says. The support page (updated August 11, 2026) says “government-issued photo IDs from most countries.” The verification article itself does not link to a specific list, but Anthropic has a separate Supported Regions Policy page—which determines which countries the company operates in at all.

Data on supported regions in the original list as of October 2, 2026:

ListSupported countries in the sample shown
Commercial API accessKazakhstan, Armenia, Georgia, Azerbaijan, Kyrgyzstan, Uzbekistan, Tajikistan, Turkmenistan, Moldova, Turkey
Claude.aiThe same set, plus Ukraine (excluding Crimea, Donetsk, Kherson, Luhansk, and Zaporizhzhia regions)

Belarus is absent from these lists.

How should you choose an action plan?

Three plans: complete verification if you have a suitable document, migrate from Claude to an alternative (Codex, Cursor, Cline + your own API), or use Claude Team/Claude API under a legal entity on Commercial Terms. Each plan has its own price and risks.

Question 1: Do you have a Pro or Max subscription, or are you on the free plan?

The risk is lower on the free plan: Anthropic triggers verification less often, and your limits are small anyway.

If you have Pro or Max, you have already paid, and the “account closed” scenario means losing your subscription. Verification has been in effect since July 8, 2026; it is worth deciding now rather than putting it off.

Question 2: Can you set up Claude API under a legal entity (your own or a partner’s)?

API access is covered by Commercial Terms—KYC is already built into signing the contract. The behavior for these accounts does not change.

Each plan is discussed below.

Preparing your document and checking the result

Step 1: prepare a physical document.

Persona does not accept scans, screenshots, or re-photographed pictures. Only a real camera—on your phone or laptop—capturing a physical document. Do not try to submit a photo of a passport shown on a screen.

Step 2: complete verification proactively.

Anthropic launches verification selectively (“when accessing certain features”). I would try to trigger it myself—check Settings → Privacy & Security and see whether a “Verify identity” button has appeared. If it has, click it right away, while the process is relatively “soft.”

RiskWhat to check
Persona still rejects a document from another countryCheck that the document is an international passport, not a local one. Local licenses from a small country may be less likely to pass than a passport
A repeat check is triggered 2–3 months after verificationPlan A is not forever—keep Plan B as a backup even after success

If you pass verification, make a note of it somewhere. Verification has been in effect since July 8, 2026, and the trigger may come at any time; completing verification in advance gives you a head start so you can calmly decide what to do next.

Where to migrate if you cannot pass?

Key takeaway

There are five alternatives: Codex CLI, Cursor, Antigravity CLI, Cline or Aider, Qwen Code, and GLM. Antigravity CLI has been closed to individual users since June 18, 2026—Enterprise only. With Cline or Aider, you choose your own application and connect any API yourself. Qwen Code and GLM are Chinese models, but their quality is lower than Claude’s. Access conditions depend on the provider you choose.

Below is a comparison table. Detailed explanations of each route are provided in separate guides linked below.

ToolKYC / setup as of October 2, 2026Quality vs. Claude CodeWho it suits
Codex CLI—On par with Sonnet 4Those accustomed to the terminal
CursorNoStronger as an IDE, but not a CLIThose who work in an IDE, not the terminal
Antigravity CLIConnect through a Google accountSimilar to CodexEnterprise plan only—for teams
Cline + your own keyKYC with the provider, not the toolDepends on the model you connectThose willing to manage the connection themselves
Qwen Code / GLM-5.3NoWorse than Claude Sonnet, but 3–5 times cheaperThose who value price over quality

Options:

If you are a solo entrepreneur building your own project. Codex CLI through an API provider. Quality is close to Claude, and the workflow is almost the same. The transition is discussed in detail in the guide how to migrate from Claude Code to Codex in 2026.

If you work in a team and need different models for different tasks. Cline or Aider, and you decide which API to connect for each task. Cline is an open-source VS Code extension—you can connect Anthropic API (through a reseller), OpenAI, DeepSeek, and Chinese models. The Codex and Claude combination is described in how to use Claude Code and Codex together in 2026.

If your budget is tight and quality is secondary. Alibaba’s Qwen 3.7-max—three times cheaper than Claude. A separate review is in the guide Qwen 3.7-max vs Opus 4.7: how to set up a Claude Code competitor at three times lower cost.

If you want to see the overall map first. Check What to code with in 2026—it compares 8 tools in one table.

What stopped working and has not returned.

  • Antigravity CLI as a free alternative—on June 18, 2026, Google closed the free plan for individual users; as of October 2, 2026, it is still Enterprise only. Details are in the guide migrating from Gemini CLI to Antigravity CLI before June 18, 2026.
  • Third-party clients with a direct Anthropic key—if you get a key as your own user, verification is triggered on the Anthropic side based on the key. Anthropic cannot see the person themselves. This is weak protection.

Can you stay on Claude through the API without verification?

Key takeaway

Yes, through Claude Team or Claude API under a legal entity. These plans fall under Commercial Terms—KYC is done when the contract with the company is signed. No one checks the individual separately. It costs more than Pro, but falls under the commercial terms for using Claude. The main bottlenecks are payment in dollars and signing a contract in English.

Plan C works, but it is not for everyone. Conditions follow.

Claude Team:

  • From $20 per user per month with annual billing ($25 when billed monthly).
  • At least 5 users on the team.
  • Registration requires payment in dollars.
  • You need to register an organization in the system—technically, Anthropic asks for the organization’s name and a contact person.
  • After registration, all team members receive full access to Claude without personal KYC.

Claude API:

  • Pay for actual usage—you add funds to your balance, then pay for tokens.

  • Minimum top-up: $5.

  • You need to set up a developer account—there is KYC there too, but it is less strict (business details are required, not a passport).

Where Plan C breaks down: Anthropic has significantly tightened API checks recently. If you exhibit a suspicious pattern (spend a lot quickly, log in from different IPs), the account is frozen, followed by correspondence with support.

Plan C makes sense if:

  • You really have a team of 5+ people, and everyone needs Claude.
  • You have a legal entity in a country where Anthropic operates normally.
  • You are willing to pay more for stability.

This is a path for businesses, not individuals.

What about projects already made with Claude Code?

Key takeaway

Your projects stay on your computer—they will not disappear. Only the Claude Code client itself is at risk. If you do not complete verification before Anthropic triggers it for you, access to Claude Code through your subscription will be disabled. The context you have accumulated—CLAUDE.md, business/, your skills—can be preserved and adapted to Codex or Cline. Plain-text rules are easier to transfer than client-specific configuration; check the new client’s loading and behavior before relying on it.

This is an important point on its own, because many readers have weeks of work behind them inside Claude Code.

What definitely remains in any scenario:

  • All your project files on disk.
  • Your CLAUDE.md files—the text remains yours, but another client may need a different filename and adapted instructions.
  • The business/ folder (if you have a second brain)—it transfers as is.
  • Skill and subagent definitions—the files remain on disk, but supported metadata, tools, hooks, and invocation formats vary by client and must be checked separately.

What gets blocked if your account is closed:

  • The Claude Code terminal client itself (it calls the Anthropic API).
  • Your chat history on claude.ai.
  • Projects inside Claude Projects (if you created any).

How to move context to a new tool.

  1. Back up CLAUDE.md and business/

    Copy these folders somewhere in the cloud or onto a flash drive. This is your second brain; without it, any AI tool works like it does for a beginner.

  2. Install an alternative client

    Codex CLI installs with one command: npm install -g @openai/codex. Cline is a VS Code extension. Details are in the migration guides.

  3. Add an API key

    Register with the provider you choose, get a key, and enter it in the client settings.

  4. Move CLAUDE.md

    For Codex, put applicable project rules in AGENTS.md; CODEX.md is not the default project instructions filename. For Cline, the documented workspace layout uses Markdown files in .clinerules/ or .cline/rules/; use one supported directory and check the Rules panel. Reuse ordinary project context and rules, then adapt client-specific commands, hooks, subagent definitions, skill metadata, and settings separately. Check Codex’s instruction discovery rather than assuming every saved file is automatically loaded.

  5. Run one task

    Take a small task and run it in the new client. Make sure the result is on par with what Claude produced. If not, adjust the prompt or change the model.

In the experience described, moving from Claude Code to Codex takes about an hour for an average project. Do not try to do everything on the day your account is closed. It is better to rehearse in advance so you can see where the seams are.

What does Persona do with your data?

Key takeaway

Persona Identities is a US identity verification service. Data is processed under US law. Your passport is stored by them, not Anthropic, is encrypted, and is not used to train models.

The legal side, briefly.

What this means in practice:

  • Your passport and selfie go to Persona Identities’ servers (the company is registered in the US).
  • Anthropic receives only the verification result (“yes/no”), not the document itself.
  • Persona has not publicly disclosed how long it retains data.
  • If you have questions about deletion, the request goes to Persona, not Anthropic.

The argument against trusting it. On February 19–20, 2026, Persona suffered a breach—about 2,500 frontend code files (53 MB). In other words, there is a recent precedent for a compromise of this very service.

What to consider when deciding. I would not give a passport to a service with a public breach in its history unless absolutely necessary. If Plans B and C are realistic for you, pursue them. Plan A, verification, is a decision to make when there is no choice.

What next?

Key takeaway

It is better to decide now rather than put it off: verification has been in effect since July 8, 2026, and the trigger comes without warning. The safest plan is to build a system where switching clients breaks nothing. That is the point of systematic work with AI: your second brain and context engineering move with you between tools in an hour. Practical materials: Ailoot documentation.

The takeaway from this situation is simple.

No one can guarantee that access to Claude will not change in a month. And Anthropic makes no promises—they do not say exactly whom they will check or when.

What you can do is prepare so that it does not matter. Move context out of a specific client and into your own system. Put your working prompts, methodologies, and business/ folder somewhere you can copy them into a new tool in an hour. Have a backup plan you have already rehearsed in real life. A plan on paper does not work.

Three practical steps:

  1. Back up your context. CLAUDE.md, business/, skills, and working prompts—all of it belongs in git. If you do not have a repo yet, create one this evening.

  2. Run one task in an alternative client. Codex or Cline—it does not matter. The goal is to make sure you know how to work there before Claude is disabled.

  3. Decide on Plan C if you have a legal entity. Registering Claude Team takes 1–2 days and removes the risk for you as an individual.

Practical materials: Ailoot documentation.

Practical materials: Ailoot documentation.

More materials and tools: Ailoot.

More details

Questions about Identity Verification in Claude: Persona and Action Plans | Ailoot

Will losing access to Claude Code delete my local project?

Your local project files and saved context remain on disk. Access to the client or account is separate, so back up the local files and rehearse a small task in another client before you need to switch.

Which instruction files should I use after moving to Codex or Cline?

Use AGENTS.md for Codex project instructions. Cline supports Markdown rules in .clinerules/ or .cline/rules/ and shows detected rules in its Rules panel. Transfer ordinary rules, but check client-specific hooks, subagents, skills, and settings separately.

All articles