You just topped up Claude Code Max, opened your terminal ready to get to work, and ran straight into a cold one-liner: "Your account has been disabled after an automatic review of your recent activities." Or maybe it was "This account has been terminated for violating our usage policy." No warning, no idea what tripped the wire, and your money is still sitting in the account — plenty of developers have been through exactly this moment.
According to a transparency report Anthropic published in January 2026, the company banned 1.45 million accounts in the second half of 2025, with an appeal success rate of just 3.3% (1,700 successes out of 52,000 appeals). A fresh wave of bans in March 2026 once again hit mainland Chinese developers hard — getting banned is now a standing risk of using Claude Code from that region.
This guide rounds up 20+ real-world reports and organizes them into three layers: understanding why you got banned, how to appeal in a way that actually has a shot, and how to keep working if the appeal goes nowhere.
Table of Contents
1. 5 Common Reasons for Bans
Anthropic's risk-control system is fully automated, so understanding what triggers it is the first step to everything else.
1. Abnormal IP activity: frequent node switching / datacenter IPs
This is the #1 reason developers outside the US get banned. Proxy nodes (datacenter IPs) are extremely easy for Anthropic to flag as non-human traffic; hopping between nodes frequently (Los Angeles one minute, Tokyo the next, then Seoul) reads to the risk-control system as classic "account compromised" or "shared account" behavior, and triggers an automatic ban almost immediately.
2. Payment risk signals: virtual cards / billing address mismatched with IP
Topping up with a virtual credit card from an unclear source, a card-reload platform, prepaid cards, or crypto-to-card services all count as high-risk payment sources. When a large charge — $100 or $200 — comes in from a high-risk payment source, the account is often frozen within minutes of the top-up. A billing address country that seriously mismatches your current login IP country will also trigger payment risk controls.
3. Using unofficial clients (the biggest offender in 2026)
Starting in January 2026, Anthropic rolled out technical countermeasures specifically targeting third-party apps that access Claude through Claude Pro/Max subscription credentials — OpenCode, Cline, and openai-proxy-style tools were hit hardest. These tools relay subscription traffic in a man-in-the-middle fashion, and Anthropic can identify the pattern from request fingerprints; accounts using this class of tool were banned en masse in the March wave.
4. Usage policy violations (content violations)
Generating harmful content, conversations that attempt to bypass safety guardrails, or highly repetitive automated requests all trigger content-based risk controls. Severe cases result in an immediate permanent ban, with almost no chance of a successful appeal.
5. Account sharing / seat-splitting
Multiple people sharing a single Claude account (especially a paid subscription) leads to anomalous-login detection (the same account logging in from different geographic locations within a short window). The system treats this as an account-compromise risk and triggers a security ban.
⚠️ On getting banned right after topping up Claude Code Max
This scenario is especially common: you top up $100 or $200 to upgrade to Max, and get banned the moment it lands. Usually it's a payment risk flag (virtual card) or a historical IP record triggering a delayed ban. In this case, always attach a payment screenshot to your appeal — this is both the scenario most likely to be a false positive and the one with the relatively highest appeal success rate.
2. First, Figure Out: Suspended or Banned?
There are two kinds of bans, and they need to be handled differently — confirm which one you have first:
| Type | Message you'll see | Severity | Appeal odds |
|---|---|---|---|
| Suspended | "Your account has been suspended" | Temporary, usually reversible | Relatively high, common with false positives |
| Disabled / Banned | "Your account has been disabled" / "terminated" | Permanent, hard to reverse | Very low (3.3% overall success rate) |
How to check:
- Log into
claude.aiand check the exact wording of the message - Log into
console.anthropic.comand check whether your API key is still valid - If the API key still works (only the web app is blocked), it's likely just a web-side violation, and API access is unaffected
3. The Complete Official Appeal Process
Channel 1: The website appeal form (recommended first choice)
- Log into
claude.aiusing the banned account (you must be logged in, otherwise you can't reach the appeal entry point) - Follow the on-page prompts to find the "Submit an appeal" button and fill out the form
- Focus on: account email, a description of your use case, and why you believe this is a mistaken ban
Channel 2: Email appeal
- Claude.ai regular users:
support@anthropic.com - API developers / Console users:
api-support@anthropic.com
Response time
Normally 3–7 business days; during peak ban waves (like March 2026) this can stretch to 2–3 weeks. It's best to send your appeal within 24 hours of being banned, while your memory of your own usage is freshest and your evidence is most complete.
4. Appeal Email Templates
Below is an effective template synthesized from multiple real-world reports. The key is to clearly explain: who you are, what you use Claude for, why you believe this is a mistaken ban, and what outcome you want.
Subject: Appeal for Account Suspension - [your@email.com]
Dear Anthropic Support Team,
I am writing to appeal the suspension of my account ([your@email.com]).
My account was suspended on [date], and I believe this may be an error.
I use Claude Code exclusively for [describe your legitimate use case —
e.g., "personal software development projects / code review / research"].
I have not knowingly violated Anthropic's usage policies. My typical
usage includes [briefly describe 3-5 of your last use cases, e.g.,
"writing Python scripts, debugging TypeScript code, drafting documentation"].
I am attaching payment confirmation and any other relevant information
that may assist your review.
Requested outcome: Reinstatement of my account.
Account email: [your@email.com]
Subscription type: [Claude Pro / Max / Free]
Thank you for your time and consideration.
Best regards,
[Your name]Subject: Account Appeal - [your@email.com]
Hello,
I am appealing the decision to ban my account ([your@email.com]).
My account was suspended on [date], which I believe may be a
system error. My primary use case for Claude Code is: [e.g.,
"personal software development, code debugging, technical writing"].
I have not intentionally violated Anthropic's usage policies. My
recent typical usage includes: [list 3-5 real use cases, e.g.,
"writing Python scripts, debugging TypeScript code"].
Attached is payment confirmation and related screenshots for your
review.
Requested outcome: Restore my account access.
Account email: [your@email.com]
Subscription type: [Claude Pro / Max / Free]
Thank you for your review — I look forward to your response.
Best regards,
[Your name]5. Key Tips to Improve Your Success Rate
What noticeably improves your odds
- Attach payment screenshots: if you were banned right after a top-up, a credit card statement screenshot is your strongest evidence (it shows you're a normal paying user)
- Describe your use case specifically: don't say "I was using it normally" — say "I was using Claude Code to debug TypeScript type errors in a React project"
- Recall and list your last 5–10 conversations before the ban: give Anthropic's reviewer concrete evidence that your usage was legitimate
- Be sincere, don't question Anthropic's judgment: don't accuse them or threaten legal action
- Send the email from a clean network: don't send your appeal from the same IP that triggered the ban
- Appeal within 24 hours: the faster you appeal, the fresher your evidence, the higher your odds
What lowers your odds
- Vague claims ("I didn't violate anything") with no concrete evidence
- An emotional, complaining, or threatening tone
- Sending multiple identical emails repeatedly (this only pushes you further back in the queue)
- Appealing from the same network environment tied to the banned account
6. Keep Working After a Ban: API Relay Options
With only a 3.3% success rate, an appeal is, in most cases, a long wait that's likely to end in failure. The more practical move is to switch immediately to an API relay that doesn't depend on your Anthropic account while your appeal is pending, so your workflow keeps running.
What is an API relay?
An API relay station is a third-party service that holds official Anthropic API quota and exposes it to developers through an OpenAI-compatible or native Anthropic-format interface. All you need is a relay API key —
point ANTHROPIC_BASE_URL in Claude Code at the relay's address, and you can keep using all of Claude Code's features — with no need for your own Anthropic account at all.
How to configure it (using Claude Code as an example)
# Add these to ~/.zshrc or ~/.bashrc:
export ANTHROPIC_BASE_URL="https://your-relay-domain.com/v1"
export ANTHROPIC_API_KEY="your-relay-API-key"
# Reload the config
source ~/.zshrcANTHROPIC_BASE_URL="https://your-relay-domain.com/v1" \
ANTHROPIC_API_KEY="sk-xxx" \
claude# After installing CC Switch, pick your relay Provider in the GUI
# Enter the base_url and API key, apply, no manual config editing needed
# See: /en/blog/cc-switch-install-guide-2026How to pick a relay
When choosing a relay, focus on three things: whether it supports Claude Opus 4.8 / Sonnet 4.6 (the models Claude Code calls by default), whether it offers a direct connection without needing a proxy, and whether pricing is transparent. Our AI API relay comparison page keeps track of ratings, pricing, model coverage, and connectivity for popular relay providers, and is worth checking as needed.
- SiliconFlow: direct connect, rated 4.2, open-weight models free, Claude models billed by usage — good for individual developers
- OpenRouter: a leading international relay, rated 4.2, widest model coverage, requires a proxy to access
- 302.AI: direct connect, rated 3.6, has a free trial tier
- PoloAPI: direct connect, rated 4.0, gives new users roughly $3 in free credit, transparent Claude 4.x pricing
- KoalaAPI: rated 3.6, minimum top-up around $1.40, failed requests aren't billed
Basic risk hygiene when picking a relay
- Top up no more than a week's worth of usage at a time (limits your losses if the provider disappears)
- Prefer providers that are company-run and properly registered
- Test latency and real model quality with a small top-up before committing
- Check the "last verified" timestamp on our comparison page, and switch away promptly from any provider that's gone offline
7. Prevention: Avoid Getting Banned Again
Network layer
- Use a fixed node: pick one stable residential IP proxy and don't switch servers often — keep the same exit IP for the entire Claude Code session
- Avoid datacenter IPs: datacenter IP ranges (AWS, GCP, Alibaba Cloud, etc.) are extremely easy to flag as bot traffic
- Don't switch proxies mid-session: switching IPs within a single session is a high-risk action
Payment layer
- Use a real credit card, with a billing address country that matches your usual login IP country
- Avoid card-reload platforms, gift cards, prepaid cards, and crypto-to-card services
- Top up smaller amounts more frequently, no more than what your typical usage requires
Usage layer
- Only use official clients: the official Claude Code CLI is the safest option; since 2026, third-party tools like OpenCode and Cline that relay traffic through subscription credentials have been a major target for Anthropic — using them significantly raises your ban risk
- Don't share accounts
- Avoid a burst of high-frequency requests in a short period (bulk automated scripts)
- Don't log into the same account frequently from different devices
The ultimate fix: use an API key directly, skip the subscription account
For heavy Claude Code users, the most thorough way to avoid bans is to never depend on an Anthropic subscription account in the first place — use an API key (official API or a reliable relay) with Claude Code instead, which eliminates the risk of account bans at the root. API usage is billed and can be precisely controlled, and for heavy-usage scenarios, the cost is often lower than a Max subscription anyway.
Summary
If you've been banned, follow this order
- Confirm the ban type (suspended vs. disabled)
- Immediately configure a relay API key so Claude Code keeps working
- Send your appeal email within 24 hours, with a payment screenshot and a specific description of your use case
- Wait 3–7 business days, don't send repeat emails
- If the appeal fails: keep using the relay option, deactivate the old account, and use a real credit card + a fixed residential IP when re-registering
Getting banned is painful, but a relay option keeps it from becoming a full work stoppage. Maintaining a stable backup path (even if it's just a spare API key on hand) is now a must-have for developers outside the US using Claude Code in 2026.
Our AI API relay comparison page keeps tracking the latest status of major relay providers — a good place to start if you're looking for a suitable alternative right now.