Anthropic

I purchased an annual $200 subscription to the "Claude Pro" AI service on September 22,2025, for professional novel editing. Shortly after purchase, and during the trial period I experienced critical, prolonged technical failures, including major file upload issues (converting. Docx to unreadable. Png files), constant crashes, inability to access the service, and a fundamental failure by the AI to retain essential editing instructions across chats.

I wasted so much time troubleshooting these bugs, asking for help, and dealing with very restrictive limitations with the premature exhaustion of paid usage tokens, effectively punishing me for their product defects.
Anthropic's support was significantly delayed, and even after acknowledging there were issues (see pic of email) they refused any reimbursement, citing an inflexible 14-day refund policy and basically steeling my money, despite documented usage of only approximately two weeks of semi-functional service since date of subscription. This amounts to charging for a non-existent service and is a deceptive practice equaling a "dark pattern" scheme, as the product failed to deliver the value of an annual subscription.
Anthropic's business practices are questionable at best. They charged full annual fees for services that are critically defective, are negligent in assisting with troubleshooting their glitches, token uses are restrictive, and they refuse to refund based on the actual usage. See list of issues below.
O The system wasn't available, messaging showed that the service was unavailable ("Claude will return soon").
O Failure to retain complex editing instructions (wasting 10-20 minutes per use).
O Can't upload a full chapter, can only upload sections of material.
O Interferes with work flow with incorrect results.
O A bug converted uploaded. Docx files to unreadable. Png format.
O Reached max usage/session limit due to troubleshooting bugs—wasted "tokens" and my time.
O I emailed multiple times regarding crashes, access, bugs/glitches and using tokens for troubleshooting without a response.
O I tried to cancel they responded by suggesting I upgrade to an even more expensive plan.
O Anthropic Support email acknowledged the delay in response times due to a high volume of inquiries, but continued their refusal of a refund.
O Anthropic refused to refund the $200 even though I'd only used it for a few weeks and most of that troubleshooting issues. I paid for a full year! They'd wasted my time with their faulty AI, ignored multiple requests for assistance, didn't provide adequate help, and wouldn't even prorate the non-use portion of the year subscription. They cited their 14-day policy.
O They admitted the situation is "not fair" (Dec 2,2025) but maintain the inflexible 14-day policy.
O Anthropic made false claims about usage being three months: Sept, Oct, and Nov, but failed to acknowledge or distinguish between use and time spent troubleshooting their malfunctioning AI, and/or waiting for customer support or assistance.
O These are deceptive practices called "dark pattern" or "roach motel"
The subscription was for a full year, but the service was critically flawed and unusable for the stated purpose for the short time use was attempted. Anthropic's refusal to refund or even prorate the unused portion of the annual fee paid, shows their unfair and deceptive business practice and constitutes a failure to uphold their end of the service agreement.