The 'Instant Ban' Phenomenon: Navigating Claude's Aggressive Fraud Detection

Recent reports from the developer community on Hacker News reveal a frustrating pattern: users attempting to subscribe to Claude Pro are being banned almost immediately after their payment is processed. In some instances, users receive an invoice via email and a Terms of Service (ToS) violation notice within the same minute, leaving them with a paid subscription they cannot access.

This phenomenon highlights a critical tension in the current AI landscape: the struggle between scaling a consumer product and implementing aggressive automated fraud detection to combat botting and token reselling.

The Mechanics of the 'Instant Ban'

For many affected users, the experience is identical. The process typically follows this sequence: account creation, credit card submission, payment confirmation, and an immediate logout followed by a ban notification.

One user, @AnthropicWHAT, noted the speed of the automation, observing that the receipt email arrived after the ban notification, underscoring that the system is triggering suspensions based on payment or account metadata before the user even has a chance to interact with the model.

Potential Triggers for Automated Suspensions

While Anthropic has not provided public documentation on their specific fraud triggers, community discussions suggest several likely culprits:

1. Payment Method Flagging

Some users found that specific credit cards trigger automatic bans. One user reported that creating a new account with the same personal information but a different credit card resolved the issue, though they noted a significant downside: the original banned account was still charged.

2. Geographic and Regulatory Constraints

As with most US-based AI companies, sanctions and regional restrictions play a role. Users in sanctioned countries or those using certain banking institutions may be flagged by automated risk-assessment tools.

3. Bot and Reseller Prevention

There is a significant industry-wide effort to stop "model distillation" (using a high-end model to train a smaller one) and the unauthorized reselling of tokens. As user @seanhandley pointed out, these activities lead companies to enable highly sensitive automated checks to keep bots out, which unfortunately results in "false positives" for legitimate human users.

User Friction and Systemic Issues

Beyond the immediate ban, users have reported several systemic frustrations with the account management system:

• Identity Verification: Some users report being forced into "Persona ID checks" to regain access or even to unsubscribe, creating a privacy barrier for those unwilling to provide government identification.
• Billing Rigidity: Reports indicate difficulties in managing subscriptions, with some users finding they can only resubscribe via yearly plans rather than monthly options after a certain period.
• Lack of Recourse: Because the bans are automated, users often find themselves in a loop of automated emails. One user reported that after multiple failed attempts to resolve the issue via email, they were forced to perform a chargeback, which likely permanently blacklisted their payment method.

Conclusion

The experience of being banned seconds after paying for a service is a textbook example of the "false positive" problem in automated security. While the need to prevent fraud and abuse is legitimate, the current implementation appears to be creating a hostile onboarding experience for a segment of legitimate customers. For users encountering this, the community suggests that public visibility—such as posting on platforms like Hacker News—may be the only way to get a human reviewer to intervene.