A major glitch in Anthropic's payment system resulted in a user on the free tier from South Korea receiving a bill for an astronomical amount — over $16.6 million. The incident, which occurred in early July, exposed serious vulnerabilities in automated billing processes that could affect any user of AI services.
How the "phantom" bill grew tenfold in a day
A developer under the pseudonym remy_notes, using Anthropic's free tier, received an initial bill for $1.67 million on July 7. The very next day, the amount increased tenfold — to $16,627,739. The user had not performed any operations, connected an API, or linked a payment card to their account. In their personal dashboard, API usage metrics remained at zero, and no paid keys were present.
Initially, remy_notes suspected phishing, but the emails came from Anthropic's official domain via the Stripe payment service. The bank rejected two attempts to debit the funds on July 8, as the amounts exceeded the card limit, but after several attempts, the user's primary card was ultimately blocked.
Official explanation and hidden risks
Anthropic acknowledged the error and assured that no actual funds were debited. The cause was attributed to a glitch in the settings of the auto-recharge credit feature — this option has since been disabled for security purposes. However, company representatives did not explain why the auto-recharge amount reached such absurd values. Importantly, no unauthorized access to the service was recorded.
Resolving the issue took four days and required approximately 18 emails from remy_notes to obtain written confirmation of the invoice cancellations. Instead of a clear response, they received automated replies, which forced them to set a deadline for Anthropic.
A systemic problem: million-dollar overcharges at major corporations
This case is just the tip of the iceberg. Startup Vaudit audited $34 million in AI service invoices for 60 large clients and identified overcharges of approximately $1.7 million — about 5% of the total amount. The largest errors are specifically in AI solution billing. Among the affected companies are Panasonic, HP, and Honda.
Analyst's opinion: Incidents like this demonstrate that the artificial intelligence industry, despite its technological leadership, is catastrophically lagging in financial control and payment system security. For users, this is a clear signal: regularly check your payment panels, monitor bank notifications, and respond promptly to any suspicious charges. One random glitch can turn into a serious financial problem, even if the money is not formally debited.