Claude account banned: how to appeal and keep working

Why Anthropic bans or suspends Claude accounts

You open your inbox and find a "your account has been suspended" email from Anthropic. You are not alone. Through 2025 and into 2026, thousands of developers and everyday users have reported sudden account restrictions, sometimes with no clear explanation. Once you know the reason, it is easier to get access back or to find a reliable alternative.

Terms of Service violations

Anthropic's Acceptable Use Policy is broad. These actions trigger enforcement:

• Prohibited content. Attempts to produce malware, weapons instructions, CSAM or material that impersonates real people can lead to an immediate permanent ban.
• Automated abuse. Sending thousands of rapid requests through the web UI (as opposed to the API), or using browser automation to get around rate limits, violates the ToS.
• Account sharing. One Pro or Teams subscription shared by several people who are not on the same billing plan counts as a violation.
• Reselling access. Using your API key to give third parties access to Claude without going through Anthropic's partner program is explicitly prohibited.

Enforcement is often automated. Anthropic's trust-and-safety systems flag accounts by pattern, so false positives happen regularly.

Region restrictions

Anthropic does not serve every country. Its products are available only in the regions on a published list, and Russia is not on it. What the policy says and what is at stake on 12 November 2026 is covered in the next section.

If Anthropic decides you are connecting from a restricted region, your account can be suspended without warning, even if you signed up from an allowed country.

VPN and proxy detection

This is one of the most common causes of sudden bans. Plenty of developers use a VPN for good reasons: privacy, company rules, remote work. But Anthropic actively detects VPN and proxy traffic. If your IP suddenly jumps to a datacenter range tied to VPN providers, the system may put your account under review or suspend it outright.

The frustrating part is that everything can work for months, and then a single IP rotation ends it. In most cases there is no grace period and no warning.

Suspicious activity patterns

Anthropic watches how you use the service:

• Rapid context switching. Dozens of new conversations per minute.
• Token volume spikes. Going from 10,000 tokens a day to 2,000,000 in one session.
• Parallel sessions. Many concurrent conversations that look automated.
• Credit card fraud signals. Prepaid cards, cards from countries other than your account region, or cards linked to chargebacks on other platforms.

Payment and billing issues

Failed payments are a common cause that people tend to underestimate:

• Expired card. If the card on file expires and the renewal charge fails, Anthropic may suspend API access within 48 hours.
• Insufficient funds. Repeated failed charges can trigger a permanent billing block.
• Disputed charges. A chargeback through your bank will almost certainly get the account banned for good.
• Sanctions screening. Your bank or payment processor may block transactions to Anthropic, and the payment errors then cascade.

Anthropic's supported regions policy and the 12 November 2026 date

Anthropic publishes a Supported Regions Policy page with an effective date of 12 November 2026. We opened the page and the policy PDF on 2 November 2026 and summarise them close to the text.

What it says:

• Anthropic's products and services are available only in the listed countries and regions. That covers its apps (Claude.ai, Claude Code, Cowork), the API and developer platform, and Claude models offered through cloud platforms and authorized resellers.
• Any country or region that is not listed is unsupported. Territories and dependencies of a listed country are also supported.
• Use by people physically located in an unsupported region counts as unsupported, including users and personnel of otherwise supported entities.
• Use by entities incorporated or headquartered in an unsupported region, and by their users or personnel, counts as unsupported even if those people sit in a supported country.
• The same goes for entities that are majority-owned or controlled, directly or indirectly, by persons or entities in unsupported regions.
• The list has 185 countries and regions. We did not find Russia, Belarus, China, Hong Kong, Iran, Syria, Cuba or North Korea on it. Ukraine is listed with exceptions: Crimea, Donetsk, Kherson, Luhansk and Zaporizhzhia.

What the text does not say: how enforcement changes on 12 November, how the new version differs from the previous one, or how the policy treats independent API gateways.

We are not lawyers and draw no legal conclusions. Whether the policy applies to you depends on where you are physically located, where your organization is registered and who owns it. If this matters for your business, talk to a lawyer or to Anthropic. codexkey is a separate service, not part of Anthropic, and we cannot promise that access will stay the same if the vendor's rules change. A sensible minimum: do not run a single integration through one channel, and keep an eye on the policy text.

What to do before 12 November

The effective date is close, and the page does not explain what changes that day. While things are unclear, order in your access and data is what helps. The step-by-step material:

• Does Claude work in Russia: a table by product, a retelling of the policy and a list of what the page leaves unsaid.
• What to do before November 12: an evening checklist, from finding hardcoded keys to testing a fallback channel.
• Exporting your Claude data: how to export chat history while the account still works, and what is left after a ban.
• Moving a project to a gateway: address, key, model names, checks and rollback.
• Claude not working: how to tell an Anthropic outage from a problem with your key, account or region.
• Gemini CLI in 2026: what became of it, if you were looking for a Claude Code replacement.

Types of account restrictions

Restrictions differ. Here is what each one means in practice.

Temporary suspension

What it looks like: you see "Your account has been temporarily suspended" when logging into claude.ai, or your API calls return 403 Forbidden with an error body that mentions account status.

Duration: typically 24 hours to 14 days. Anthropic's support documentation says temporary suspensions are lifted automatically after review, but in practice many users report waiting 2-4 weeks.

Can you recover? Yes, this is the most recoverable type. It is usually triggered by automated fraud detection or a sudden usage spike.

Permanent ban

What it looks like: an email from Anthropic saying your account has been "terminated" or "permanently disabled". API keys stop working immediately, and you cannot log into claude.ai.

Cause: repeated ToS violations, confirmed abuse or a chargeback dispute.

Can you recover? Rarely. An appeals process exists, but the success rate for permanent bans is low, estimated at <15% based on community reports.

Region block

What it looks like: you can log in, but all API calls and conversations return errors related to your geographic location. Sometimes the web UI loads but refuses to send messages.

Cause: IP geolocation places you in a restricted country.

Can you recover? Yes, if you really are in an allowed region and were flagged by mistake. You will need to provide proof of residence.

Rate limit freeze

What it looks like: the account works, but you are permanently stuck at the lowest tier of rate limits. Upgrading your plan or adding a card does not change the limits.

Cause: earlier abuse or suspicious activity that did not justify a full ban but led to a permanent rate reduction.

Can you recover? This is the hardest restriction to get lifted, because Anthropic often does not acknowledge it explicitly.

Account deletion

What it looks like: the account no longer exists. Logging in shows "account not found", and all conversation history is gone.

Cause: either you requested deletion under GDPR/CCPA, or Anthropic deleted the account as part of an abuse cleanup.

Can you recover? If Anthropic deleted it, no. The data is gone. You would have to create a new account from scratch, unless you are blocked from registering.

Step-by-step recovery plan

Step 1: Check your email

Look for any message from Anthropic. A notification usually reaches your registered address within 24 hours of an account action. Check the spam folder too, because these emails often land there.

Step 2: Identify the restriction type

Try logging into claude.ai and making an API call. Write down the exact error message and the HTTP status code. You will need both for your appeal.

Step 3: Contact Anthropic support

Send an email to support@anthropic.com with the subject line: Account Appeal: [your email address]

In the body, include:

1. Your account email address.
2. The exact error message or the restriction you are seeing.
3. A short, factual description of how you use Claude.
4. An explanation of any unusual activity, if there was some (for example, "I was traveling and used a VPN").
5. A specific request: reinstatement, a rate limit review or region verification.

What NOT to write:

• Do not be confrontational or make threats.
• Do not lie about your usage. Anthropic has full logs.
• Do not send several emails a day, because it slows processing down.

Step 4: Wait and follow up

Anthropic's typical support response times as of 2026:

If you have heard nothing in the expected window, send a single follow-up email that references your original ticket.

Step 5: Escalate if necessary

If your appeal is denied or ignored, the options are limited:

• Social media. A post on X (Twitter) tagging @AnthropicAI has occasionally prompted a faster reply, but the result is unpredictable.
• Legal request. If you believe the ban violates your rights under GDPR, CCPA or similar laws, you can submit a formal data subject access request (DSAR). Anthropic must respond within 30 days.
• A new account. After a ban this goes against Anthropic's ToS. If you are caught, the new account will be banned too. You do this at your own risk.

If you need uninterrupted access for development, do not tie a working integration to the fate of one personal account. Create a separate API key, set a limit for the project and test it with a small request first.

What to do if recovery fails

Here is the reality: many accounts are never reinstated. If you are a developer whose production systems depend on Claude, waiting weeks for an appeal is not an option. You need another route to the same models. Anthropic's regional rules still apply along the way (see the section above).

codexkey: Claude API access without an Anthropic account

codexkey is an API gateway that gives you access to Claude models (Sonnet, Opus) without a direct Anthropic account. You sign up with codexkey, get an API key and point your existing code at our endpoint. No Anthropic account and no VPN needed. Anthropic's regional rules do not go away because of that, as the section above explains.

Haiku is in the catalogue too: claude-haiku-5.5 and claude-haiku-4.5, with prices on the models page. They run on the Premium channel, and the Cheap channel has no Haiku.

What changes in your code? One line: the base_url.

Python example (anthropic SDK)

import anthropic

# Before: Direct Anthropic (requires Anthropic account)
# client = anthropic.Anthropic(api_key="sk-ant-...")

# After: codexkey (no Anthropic account needed)
client = anthropic.Anthropic(
    api_key="your-codexkey-api-key",
    base_url="https://api.codexkey.online",
)

message = client.messages.create(
    model="claude-sonnet-5",
    max_tokens=1024,
    messages=[
        {"role": "user", "content": "Explain quantum entanglement simply."}
    ],
)
print(message.content[0].text)

TypeScript example

import Anthropic from "@anthropic-ai/sdk";

// Before: Direct Anthropic
// const client = new Anthropic({ apiKey: "sk-ant-..." });

// After: codexkey
const client = new Anthropic({
  apiKey: "your-codexkey-api-key",
  baseURL: "https://api.codexkey.online",
});

const message = await client.messages.create({
  model: "claude-sonnet-5",
  max_tokens: 1024,
  messages: [
    {
      role: "user",
      content: "Explain quantum entanglement simply.",
    },
  ],
});

console.log(message.content[0].text);

Nothing else needs to change. The application logic (prompts, tools, streaming, function calling) works as before. The SDK does the rest, and you only swap the endpoint.

With an OpenAI-compatible SDK

If your code already uses the OpenAI SDK, codexkey supports the OpenAI-compatible endpoint too:

from openai import OpenAI

client = OpenAI(
    api_key="your-codexkey-api-key",
    base_url="https://api.codexkey.online/v1",
)

response = client.chat.completions.create(
    model="claude-sonnet-5",
    messages=[
        {"role": "user", "content": "Hello from OpenAI SDK!"}
    ],
)
print(response.choices[0].message.content)

Comparison: Anthropic direct, codexkey, OpenAI and Together.ai

How to avoid account problems in the future

Whether you got your account back or moved to codexkey, these habits lower the risk of a repeat:

1. Keep your IP stable. Do not switch VPN servers while you work with Claude. If you cannot do without a VPN, pick one server in an allowed region and leave it alone.
2. Watch your billing. Set up auto-pay with a card that will not expire soon, and do not ignore failed-charge notifications.
3. Stay within published limits. Do not try to get around rate limits with several accounts or with parallel sessions beyond your tier.
4. Read the ToS. Anthropic updates its Acceptable Use Policy from time to time. Major changes usually come with 30 days of notice.
5. Keep a backup. Even if your main access goes through Anthropic directly, keep a configured codexkey API key in reserve. If the primary channel goes down, switching takes seconds.

FAQ

Can I create a new Anthropic account after being banned?

Technically, Anthropic's ToS prohibits creating new accounts after a ban. In practice Anthropic matches accounts by email, phone number and payment method. If all three are different, a new account may work, but you risk being detected again and banned a second time. A safer route is an API gateway like codexkey, which does not need an Anthropic account at all.

How long does an Anthropic account suspension typically last?

Temporary suspensions usually last from 24 hours to 14 days, but many users report waiting 2-4 weeks before access is fully restored. The timeline depends on the reason for the suspension and on how quickly Anthropic's trust-and-safety team gets to your case.

Will using a VPN get my Claude account banned?

It can. Anthropic actively detects VPN and datacenter IPs. Many people use a VPN without trouble, but a sudden IP change, especially to a region with a higher fraud rate, can trigger a suspension. If you cannot do without a VPN, use a residential IP service instead of a standard datacenter VPN.

Does codexkey support all Claude models, including Claude Opus?

Yes. codexkey gives you access to Claude Sonnet, Claude Opus and Claude Haiku (claude-haiku-5.5, claude-haiku-4.5), plus GPT, Gemini and other models. For the Claude models you use the same identifiers (for example claude-sonnet-5) as in the Anthropic SDK.

Can I use my existing code with codexkey without rewriting it?

Yes. Almost always the only change is setting base_url (Python) or baseURL (TypeScript) to https://api.codexkey.online for the Anthropic SDK. With the OpenAI SDK the address is set the same way but with a suffix: https://api.codexkey.online/v1. Streaming, tool use, function calling, vision and all other features work identically.

Is my data safe when I use codexkey instead of Anthropic directly?

codexkey acts as an API gateway: your prompts and responses pass through it on the way to the Claude models. For details on data handling, retention and encryption, read the privacy policy. As with any third-party provider, judge whether that handling meets your compliance requirements.

What payment methods does codexkey accept?

codexkey offers more payment methods than Anthropic directly, including international credit and debit cards. The current list is on the pricing page.

Can Anthropic ban me for using a third-party API gateway?

Anthropic's terms govern the relationship between Anthropic and its customers. When you work through codexkey, your agreement is with codexkey and you have no Anthropic account, so there is nothing to ban there. How Anthropic applies its regional policy to intermediary services and their customers is not stated in the policy text, and we do not try to predict it. No service can give guarantees about the future.

What changes on 12 November 2026 in Anthropic's supported regions policy?

The Supported Regions Policy page shows an effective date of 12 November 2026. By its text, Anthropic's products are available only in the listed countries and regions, and use by people physically located in an unsupported region, or by organizations registered, headquartered or controlled from one, counts as unsupported. Russia is not on the list. The page does not explain how enforcement will change after that date.

What should I do before 12 November 2026?

Find where your project hardcodes Anthropic's address and key, back up your Claude Code configuration, export your chat data and test a fallback channel with a one-token request. The full list with commands is in the checklist, and what is known about the policy itself is in Does Claude work in Russia.

How do I get my data out of a blocked Claude account?

According to Anthropic's help pages, you can still sign in to a blocked account on claude.ai to export your data or delete the account. The scope of the export may be limited depending on the violation. The steps and limits are described in Exporting your Claude data.

Does Anthropic's policy apply to people who use an API gateway?

The policy text does not mention API gateways separately. It speaks about using Anthropic's products in unsupported regions. We give no legal conclusions: how the rules apply to you depends on where you are and where your organization is registered, so check with a lawyer or with Anthropic.

Conclusion

Losing access to a Claude account is stressful, especially when production depends on it. An appeal is possible, but it is slow and the outcome is unpredictable. The better approach works on two fronts: appeal the ban if you believe it was a mistake, and set up backup access in parallel so your work does not stall.

codexkey offers that alternative: one line of code, no Anthropic account, the same Claude models you already build with. It does not cancel Anthropic's regional policy, and we give no guarantees about the future. Sign up, get an API key and keep it as a reserve next to your main channel.

Get started with codexkey →