Why did Anthropic charge a free user $16.6 million despite zero API usage?

A Korean Claude user has alleged that Anthropic tried to charge a hefty $16.6 million amount for using API and the most surprising part is that the user had zero usage.

As per report published by International Cyber Digest, the user received two massive and unexpected invoices via Stripe, supposedly from Anthropic. The first invoice for $1,669,875.90, which increased tenfold to $16,627,739.70 within 24 hours.

At first, the user thought about a sophisticated phishing attempt or an elaborate scam. But the most horrifying part is that the invoice emails came from Anthropic’s official domain and used the company’s legitimate Stripe invoicing system.

Here comes the most surprising part, the affected user states they were on a free plan and had no payment method on file. Moreover, the dashboard showed zero API usage.

Despite not having a card registered, the user’s bank recorded and declined two attempted overseas charges from the US-based artificial intelligence company on July 8. All attempted charges were declined by the bank, and no actual funds were successfully transferred.

Anthropic has not yet responded to this incident and clarified whether the incident was caused by security breach, AI hallucination or technical billing error.

The incident also took the internet by storm, raising concerns in the tech community. One user wrote, “Anthropic is really out here cold, invoicing customers in bulk hoping people just pay for it, the absolute state of reality.”

Another quipped, “Well I guess that Claude was generating the invoices. LLMs do hallucinate sometimes.”

Some called it as a phishing attempt, “This claim needs verification before anyone assumes anthropic actually billed a free plan user millions. sounds more like a phishing attempt than a real invoice.”