Technology & Gadgets | July 2026 | 4 min read
⚡ Quick Takeaways
- A South Korean Claude user reported invoices rising from $1.67 million to $16.6 million in less than a day.
- The user said his account was on a free plan and showed zero billable API usage.
- Anthropic reportedly told the user an incorrect automatic credit reload setting caused the invalid payment requests.
- The payment processor attempted collection, but the user’s bank declined the transactions and no money was taken.
- Anthropic’s public status page listed technical incidents around those dates but no billing-related incident.
Anthropic has acknowledged a billing error after a Claude user in South Korea said the company attempted to collect a $16.6 million payment despite his account showing no billable API usage.
| $16.6M
Largest invoice amount |
$1.67M
Earlier failed invoice |
18 emails
User’s reported support chase |
User First Suspected a Scam
The user said he first received a failed payment invoice for $1,669,875.90 on July 7. Less than 24 hours later, a second invoice reportedly raised the demand to $16,627,739.70, an amount so unusual that the user initially suspected phishing or a payment scam.
He later said the emails appeared to come through Anthropic’s official billing system and included payment links connected to Stripe infrastructure. Screenshots shared online also appeared to show overseas transaction attempts from a US merchant identified as Anthropic.
| Key concern: The reported invoices were not merely alarming emails; the user’s bank allegedly received attempted card transactions linked to the invalid billing requests. |
Anthropic Blames Credit Reload Setting
Anthropic later told the user that an incorrect automatic credit reload setting had generated the invalid payment requests. The company said it disabled the setting as a precaution, restored the account’s billing configuration and found no evidence that the account had been accessed without permission.
According to the user’s account, Anthropic also confirmed that the payment processor attempted to collect the invalid amount, but the transaction was declined. The user was told he did not owe the company any money. Anthropic has not publicly explained how the reload value reached more than $16 million.
| Item | Reported detail | Status | Reader takeaway |
| First invoice | $1,669,875.90 | Failed payment | Unusually large billing trigger |
| Second invoice | $16,627,739.70 | Declined by bank | No funds transferred |
| Account usage | User reported zero API usage | Disputed bill | Mismatch with dashboard |
| Company response | Incorrect auto reload setting | Billing restored | No unauthorised access found |
Support Delays Added to the Disruption
The user said the financial scare quickly became a support problem. In follow-up posts, he claimed he spent four days trying to obtain written confirmation that the invoices had been cancelled and his account had been cleared. He said he sent around 18 emails before a human support response arrived.
He also said the attempted payments led his main bank card to be blocked. Even though the transactions failed, the incident shows how a billing configuration error can create real disruption before a support team has reviewed the case.
| Customer impact: No money was reportedly collected, but the failed attempts still caused card disruption and a days-long effort to get written clearance. |
No Public Billing Incident Was Listed
Anthropic’s public service-status page did not list a billing-related incident for July 7 or July 8. The page recorded several technical problems affecting Claude models and related services during those dates, but none referred to incorrect invoices, payment attempts or automatic credit reloads.
That gap is one reason the case drew attention online. For AI tools with pay-as-you-go billing, users expect automatic controls to stop abnormal charges before they reach a card network.
| Bottom line: The case appears to be a billing-control failure, not a successful charge or an account takeover, based on the user’s reported correspondence with Anthropic. |
Billing Controls Face New Scrutiny
The broader concern is not only the size of this invoice but the path it took through automated systems. A reload setting that can generate a multimillion-dollar payment request suggests companies need stronger caps, anomaly checks and manual review before extreme billing events reach customers or banks.
For Anthropic, the episode comes as Claude is used by more developers, teams and casual users. As AI services become part of daily work, billing transparency and rapid human support are becoming as important to trust as model performance.
| Risk note: The incident is based largely on user-shared screenshots and reported support replies, so unconfirmed technical details should be treated with caution until Anthropic gives a public explanation. |
Disclaimer: This article is based on public reporting, user-shared posts and Anthropic status-page information available at the time of writing. Anthropic has not published a detailed public incident report for the billing error.
