Anthropic
Five Unexplained $200 Charges and Account Closure
- 10-21-2026
- 3
On October 9, I received a notification from my credit card company that Anthropic had processed five consecutive charges against my credit card, each for $200, totaling $1,000. I had no idea what these charges were for or how they had originated, so I did what any reasonable customer would do: I attempted to contact Anthropic to investigate and resolve the matter.
That decision marked the beginning of what I can only describe as a week of extraordinary frustration.
Having spent my career as a senior engineer, I am accustomed to troubleshooting problems systematically, following evidence, identifying root causes, and expecting people and organizations to take responsibility when something goes wrong. I understand that technical problems occur. What I find far more difficult to accept is an organization's apparent inability or unwillingness to investigate a problem when the customer provides the information necessary to do so.
Unfortunately, my experience with Anthropic demonstrated a remarkable failure in precisely these areas.
Four Days, Multiple AI Bots, and No Meaningful Progress
Over approximately four days, I dealt with what appeared to be at least three layers of AI-powered customer support. The exchanges generated an extensive email trail, eventually exceeding 40 messages, during which I repeatedly and politely explained the same problem, supplied the same information, and asked for assistance.
The responses followed a predictable pattern: requests for information I had already provided, explanations that did not address the issue, and instructions that led back to the beginning of the process. There was no apparent ownership of the problem, no meaningful escalation, and no evidence that anyone was conducting an actual investigation.
Eventually, after repeatedly encountering the same circular responses, my patience wore thin, and my language became more direct and colorful. That was when I received a warning that such language would not be tolerated by Anthropic.
Interestingly, there was no acknowledgment that it had taken more than 40 emails of polite correspondence before I appeared to get any meaningful attention. Nor was there any recognition that the increasingly frustrated language was a consequence of the support process itself.
I later came to believe that the supposed human response was simply another AI-generated interaction.
For the record, the language that apparently triggered their sensitivity was primarily my use of a mild expletive to describe the explanations being offered for why they could not help me. Perhaps that word trips a sensitivity threshold in their automated systems. If so, it appears their systems are better equipped to police a customer's vocabulary than to resolve a legitimate billing problem.
A Straightforward Problem That Nobody Would Investigate
The facts were relatively simple:
Anthropic had charged my credit card five times for $200 each.
My credit card company identified the charges as originating from Claude.
I provided Anthropic with a copy of the relevant credit card statement.
Anthropic repeatedly claimed it could not locate any invoices associated with the charges.
I supplied multiple email addresses that might have been associated with the account or accounts in question.
Anthropic repeatedly reported that it could find no matching invoices under those email addresses.
Each time I supplied the requested information, the process returned to the same starting point.
From an engineering perspective, this is a basic diagnostic problem. If a transaction exists and a payment processor or credit card statement can identify the merchant, there should be a process for tracing that transaction through the merchant's billing records. Even if the initial account lookup fails, that should be the starting point for further investigation, not the justification for ending the inquiry.
I repeatedly explained that Anthropic had access to the transaction information and that I had provided the supporting documentation. I asked them to identify what the charges were for, which account they were associated with, and which email address had been used.
I also repeatedly requested escalation to a real person who could investigate the matter.
Instead, I received the same responses, the same requests, and the same dead ends.
At no point did I receive a satisfactory explanation for the five charges.
Administrative Burden Shifted to the Customer
Because Anthropic failed to resolve the issue, I had to dispute each of the five charges individually with my credit card company.
That process was time-consuming and required completing multiple forms containing largely redundant information. It is difficult to understand why a customer should have to undertake this administrative burden when the merchant has been given the transaction details and an opportunity to investigate.
I have no objection to following a reasonable dispute process when one is necessary. What I object to is being forced into that process because the merchant's customer support system cannot, or will not, perform a basic billing investigation.
I had initially assumed Anthropic was a reputable company that would take a reasonable approach to resolving an apparent billing problem. My experience substantially changed that perception. Whatever its public reputation, the support process I encountered gave me little reason to believe that resolving the customer's problem was a priority.
The Situation Became Even More Absurd
After several days of this circular exchange, I finally received another response. Rather than escalating the matter to someone capable of investigating it, the AI bot suggested that I perform several actions within my account.
It then informed me that my accounts were being closed.
The contradiction was difficult to overlook. I was being instructed to access an account and perform troubleshooting steps while simultaneously being informed that the accounts were being closed, leaving me unable to follow those instructions.
The explanation for the closure was that I had violated Anthropic's policies. No specific explanation was provided that meaningfully connected the alleged violation to my original billing inquiry.
From my perspective, the timing and circumstances were deeply troubling. I had reported unexplained charges totaling $1,000, repeatedly requested assistance, and challenged the adequacy of the responses I was receiving. The result was not a resolution of the billing issue but the closure of my accounts.
I cannot speak to Anthropic's internal reasoning, but the sequence of events raises serious questions about how the company handles customers who persistently seek answers to unresolved billing problems.
An Automated Support Loop with No Way Out
Even after the account closure, the emails continued.
Anthropic's automated responses instructed me to log in to my account to check API status and review other account information. The obvious problem was that my account had been suspended or closed, and I could no longer log in.
When I explained this, the response once again directed me to log in.
At that point, the process had become a textbook example of a closed-loop failure: the automated system prescribed an action that the customer could not perform, and the system then repeated the same instruction rather than recognizing the contradiction and routing the case to someone who could resolve it.
In engineering, when a system repeatedly produces the same failed result, the proper response is not to repeat the same procedure indefinitely. It is to recognize the failure, investigate the underlying cause, and implement a corrective action.
That basic principle seemed entirely absent from this experience.
The Bottom Line
My complaint is not simply that Anthropic's AI support was frustrating. It is that the entire process appeared incapable of exercising the judgment, accountability, and problem-solving ability necessary to resolve a relatively straightforward billing dispute.
I encountered unexplained charges totaling $1,000, provided supporting documentation, supplied multiple possible account email addresses, and repeatedly requested a meaningful investigation. In return, I received circular responses, redundant requests, contradictory instructions, and ultimately an account closure without a satisfactory resolution of the original issue.
The irony is difficult to miss. Anthropic is a company known for developing advanced artificial intelligence, yet my experience with its customer support suggested a system optimized to generate responses rather than solve problems.
Automation can be an excellent tool. Properly implemented, it can improve efficiency, reduce costs, and help customers resolve routine issues quickly. But automation without an effective escalation path is not customer support. It is a barrier between the customer and the organization responsible for addressing the problem.
After a career in engineering, I have learned that problems are inevitable. What distinguishes a competent organization is how it responds when those problems arise. It investigates, follows the evidence, accepts responsibility where appropriate, and ensures that someone owns the issue until it is resolved.
That is what I expected from Anthropic. It is not what I received.
As of this writing, I am still waiting for a satisfactory explanation of the charges and a meaningful response to the original billing problem. The experience has left me questioning not only Anthropic's customer service practices but also its commitment to basic accountability when a customer raises a legitimate concern.
For a company developing some of the world's most sophisticated AI technology, the failure to provide a functioning path to human assistance is particularly disappointing.
Follow-up a week later
It has been a week. As of now, Anthropic claims (at least their AI bot claims) that they have no record whatsoever of charging my Credit Card. The Credit Card company is requesting copies of invoices for the charges before they can investigate. Tried to inform them that I have no invoices, and Anthropic can't, or is unwilling to try and help.
So I'm out $1000, got a canceled Anthropic Claude account, and still nothing from Anthropic. Guess there's not much more to do except wait to get billed some more from the company who conveniently can't trace charges.
Company: Anthropic
Country: USA
That decision marked the beginning of what I can only describe as a week of extraordinary frustration.
Having spent my career as a senior engineer, I am accustomed to troubleshooting problems systematically, following evidence, identifying root causes, and expecting people and organizations to take responsibility when something goes wrong. I understand that technical problems occur. What I find far more difficult to accept is an organization's apparent inability or unwillingness to investigate a problem when the customer provides the information necessary to do so.
Unfortunately, my experience with Anthropic demonstrated a remarkable failure in precisely these areas.
Four Days, Multiple AI Bots, and No Meaningful Progress
Over approximately four days, I dealt with what appeared to be at least three layers of AI-powered customer support. The exchanges generated an extensive email trail, eventually exceeding 40 messages, during which I repeatedly and politely explained the same problem, supplied the same information, and asked for assistance.
The responses followed a predictable pattern: requests for information I had already provided, explanations that did not address the issue, and instructions that led back to the beginning of the process. There was no apparent ownership of the problem, no meaningful escalation, and no evidence that anyone was conducting an actual investigation.
Eventually, after repeatedly encountering the same circular responses, my patience wore thin, and my language became more direct and colorful. That was when I received a warning that such language would not be tolerated by Anthropic.
Interestingly, there was no acknowledgment that it had taken more than 40 emails of polite correspondence before I appeared to get any meaningful attention. Nor was there any recognition that the increasingly frustrated language was a consequence of the support process itself.
I later came to believe that the supposed human response was simply another AI-generated interaction.
For the record, the language that apparently triggered their sensitivity was primarily my use of a mild expletive to describe the explanations being offered for why they could not help me. Perhaps that word trips a sensitivity threshold in their automated systems. If so, it appears their systems are better equipped to police a customer's vocabulary than to resolve a legitimate billing problem.
A Straightforward Problem That Nobody Would Investigate
The facts were relatively simple:
Anthropic had charged my credit card five times for $200 each.
My credit card company identified the charges as originating from Claude.
I provided Anthropic with a copy of the relevant credit card statement.
Anthropic repeatedly claimed it could not locate any invoices associated with the charges.
I supplied multiple email addresses that might have been associated with the account or accounts in question.
Anthropic repeatedly reported that it could find no matching invoices under those email addresses.
Each time I supplied the requested information, the process returned to the same starting point.
From an engineering perspective, this is a basic diagnostic problem. If a transaction exists and a payment processor or credit card statement can identify the merchant, there should be a process for tracing that transaction through the merchant's billing records. Even if the initial account lookup fails, that should be the starting point for further investigation, not the justification for ending the inquiry.
I repeatedly explained that Anthropic had access to the transaction information and that I had provided the supporting documentation. I asked them to identify what the charges were for, which account they were associated with, and which email address had been used.
I also repeatedly requested escalation to a real person who could investigate the matter.
Instead, I received the same responses, the same requests, and the same dead ends.
At no point did I receive a satisfactory explanation for the five charges.
Administrative Burden Shifted to the Customer
Because Anthropic failed to resolve the issue, I had to dispute each of the five charges individually with my credit card company.
That process was time-consuming and required completing multiple forms containing largely redundant information. It is difficult to understand why a customer should have to undertake this administrative burden when the merchant has been given the transaction details and an opportunity to investigate.
I have no objection to following a reasonable dispute process when one is necessary. What I object to is being forced into that process because the merchant's customer support system cannot, or will not, perform a basic billing investigation.
I had initially assumed Anthropic was a reputable company that would take a reasonable approach to resolving an apparent billing problem. My experience substantially changed that perception. Whatever its public reputation, the support process I encountered gave me little reason to believe that resolving the customer's problem was a priority.
The Situation Became Even More Absurd
After several days of this circular exchange, I finally received another response. Rather than escalating the matter to someone capable of investigating it, the AI bot suggested that I perform several actions within my account.
It then informed me that my accounts were being closed.
The contradiction was difficult to overlook. I was being instructed to access an account and perform troubleshooting steps while simultaneously being informed that the accounts were being closed, leaving me unable to follow those instructions.
The explanation for the closure was that I had violated Anthropic's policies. No specific explanation was provided that meaningfully connected the alleged violation to my original billing inquiry.
From my perspective, the timing and circumstances were deeply troubling. I had reported unexplained charges totaling $1,000, repeatedly requested assistance, and challenged the adequacy of the responses I was receiving. The result was not a resolution of the billing issue but the closure of my accounts.
I cannot speak to Anthropic's internal reasoning, but the sequence of events raises serious questions about how the company handles customers who persistently seek answers to unresolved billing problems.
An Automated Support Loop with No Way Out
Even after the account closure, the emails continued.
Anthropic's automated responses instructed me to log in to my account to check API status and review other account information. The obvious problem was that my account had been suspended or closed, and I could no longer log in.
When I explained this, the response once again directed me to log in.
At that point, the process had become a textbook example of a closed-loop failure: the automated system prescribed an action that the customer could not perform, and the system then repeated the same instruction rather than recognizing the contradiction and routing the case to someone who could resolve it.
In engineering, when a system repeatedly produces the same failed result, the proper response is not to repeat the same procedure indefinitely. It is to recognize the failure, investigate the underlying cause, and implement a corrective action.
That basic principle seemed entirely absent from this experience.
The Bottom Line
My complaint is not simply that Anthropic's AI support was frustrating. It is that the entire process appeared incapable of exercising the judgment, accountability, and problem-solving ability necessary to resolve a relatively straightforward billing dispute.
I encountered unexplained charges totaling $1,000, provided supporting documentation, supplied multiple possible account email addresses, and repeatedly requested a meaningful investigation. In return, I received circular responses, redundant requests, contradictory instructions, and ultimately an account closure without a satisfactory resolution of the original issue.
The irony is difficult to miss. Anthropic is a company known for developing advanced artificial intelligence, yet my experience with its customer support suggested a system optimized to generate responses rather than solve problems.
Automation can be an excellent tool. Properly implemented, it can improve efficiency, reduce costs, and help customers resolve routine issues quickly. But automation without an effective escalation path is not customer support. It is a barrier between the customer and the organization responsible for addressing the problem.
After a career in engineering, I have learned that problems are inevitable. What distinguishes a competent organization is how it responds when those problems arise. It investigates, follows the evidence, accepts responsibility where appropriate, and ensures that someone owns the issue until it is resolved.
That is what I expected from Anthropic. It is not what I received.
As of this writing, I am still waiting for a satisfactory explanation of the charges and a meaningful response to the original billing problem. The experience has left me questioning not only Anthropic's customer service practices but also its commitment to basic accountability when a customer raises a legitimate concern.
For a company developing some of the world's most sophisticated AI technology, the failure to provide a functioning path to human assistance is particularly disappointing.
Follow-up a week later
It has been a week. As of now, Anthropic claims (at least their AI bot claims) that they have no record whatsoever of charging my Credit Card. The Credit Card company is requesting copies of invoices for the charges before they can investigate. Tried to inform them that I have no invoices, and Anthropic can't, or is unwilling to try and help.
So I'm out $1000, got a canceled Anthropic Claude account, and still nothing from Anthropic. Guess there's not much more to do except wait to get billed some more from the company who conveniently can't trace charges.
Company: Anthropic
Country: USA