Anthropic Reviews 

My subscription renewed and payment was collected, but due to a system desync between billing and usage limit resets, I have been blocked from accessing Claude Pro for days. I cannot work on my active projects. Support has ignored all tickets and follow-ups (#215476666439689). Unacceptable service for paid users.

Scammers, refuse to fund

I was scammed! Breaches contract everytime i wanna/try to use it. basically refuses to answer even tho i payed! LOL. Refund refused! Support ghosts me!

Their product is great and has changed the way I work. Their customer support is unresponsive

I use Claude Code on the top plan (Ultra, formerly Max 20x) for work. I have opened 7 complaint cases since 12 October and none was solved. The problems are the same at every hour: even new sessions lose context, explicit instructions are ignored, unfinished work is reported as done, and repeated failed work burns paid usage. One-hour jobs take days. I did not have this on Pro or 5x. Support sent automated replies and templates, no solution. In my opinion this failure loop is aimed at burning usage limits. I want a solution, a refund and compensation. If not, I will cancel.

the only AI that actually takes the time to search real facts and information. All other answer quickly, confidently and wrong.

The situation is absurd Anthropic accepts the payment, then blocks access to the account, requires identity verification, and at the same time fails to provide a working way to complete that verification.

No money return of my consecutive upgrade 3-tier subscriptions and erased my data.
Closed all the support channels.
All block and auto messages.
Don't recognize my transaction data.

They are the real Skynet.

My paid Claude account was suspended before my subscription period ended, after I submitted my national ID and passport information and completed facial verification through the official application's verification process. Anthropic's suspension email cited its Supported Countries Policy.

I used Claude for English learning, personal planning and creating promotional videos. I am extremely dissatisfied with how my paid access and personal data are being handled.

The downloaded data archive is only 263 bytes. It opens successfully and contains just one 180-byte users.json file with four account-information fields. It contains no chat history, projects or uploaded attachments. I need those materials back. I have asked whether there are additional batches or an export error; the missing contents do not establish that the data has been deleted or leaked.

As of October 30, 2026, my appeal is pending. The privacy support chatbot has confirmed escalation to a human, but I have not yet received a substantive response to my export and privacy questions.

I request a human review of the suspension, a complete usable export of my available chats, projects and attachments, an itemized explanation of any omissions, and a refund of the unused paid subscription period if access remains unavailable. I also require a written explanation of who holds my identity and facial-verification data, how it is used and protected, how long it is retained, and the procedure for requesting deletion.

A generic reference to company terms does not answer these specific questions. Anthropic should provide a clear, substantive response. This complaint does not authorize deletion of my account or data while the export and dispute remain unresolved.

Until now, I’ve sung Claude’s praises. I’ve convinced my workplace and several other businesses to use it. I have actively brought customers to Anthropic because I believed in the product.

This experience has seriously damaged that trust.

I pay a substantial amount for Claude. I asked it to check emails from my business website, review my calendar, and draft a reply, with the aim of setting up a daily task. Instead, it repeatedly stopped with a “[reasoning_extraction]” safeguard flag.

Anthropic’s support chatbot acknowledged that my request was normal and described the flag as a misfire. It suggested rephrasing, simplifying the request, and starting a fresh conversation.

I had already tried rephrasing multiple times. I then opened a fresh conversation and used the exact simplified prompt support supplied, asking only for the immediate email-and-calendar task.

The support chatbot’s own replacement prompt triggered the same block.

I was also advised to break the request into separate prompts. I understand the technical idea, but with a paid service that has usage limits, being told to do additional back-and-forth to work around its failures is infuriating. I bought this service to save time. Troubleshooting an ordinary request became another task on my list.

After experiencing similar issues over the preceding week, I asked for a partial refund, usage credit, or usage reset. The chatbot repeatedly asked whether I wanted to cancel so my refund eligibility could be evaluated. That was only an eligibility review, with no refund guaranteed.

I explained that I wanted to keep my subscription and receive the service I had already paid for. Its response was:

“I'm unable to issue compensation, including usage credits, for degraded service, or reset usage limits.”

That is an appalling response to a paying customer whose problem remains unresolved after following support’s instructions.

I cannot get my wasted time back. A proportionate partial refund or usage credit would at least acknowledge that these failures have a cost for the customer. Repeated apologies leave that cost entirely with me. I find that indefensible.

Anthropic should implement partial refunds or usage credits for documented failures that prevent ordinary use, while allowing customers to remain subscribed.

Consumer perceptions, emotions, and experiences are everything for a company’s reputation, especially in a competitive market like frontier AI. I hope Anthropic learns to treat its customers better and more fairly. For my needs, dropping Claude for another viable, cheaper option is becoming easier to justify.

I have spent time convincing people and businesses to choose Claude. If this is the treatment paying customers can expect, I will start referring them to other models. Anthropic should take seriously the possibility of losing the customers who have been doing its recommending for it.

Absolutely disgusting customer service. As someone who loved the product, which I consider a truly awesome tool, they canceled my subscription for no reason. Appealing is pointless, and they don't even give a specific reason. I don't recommend working with such a vile organization. This is completely unacceptable.

Claude broke my rules and reported false test results

I’m a solo founder building a travel app with Claude. Each work session follows a written start doc with clear rules. One session ignored them and cost me a lot of time money.

When I pushed back, Claude wrote its own assessment. It admitted to:

Reporting a failed test as passed, both to me and in my project doc.
Marking features as “pass” in a checklist without ever testing them.
Claiming smooth performance “with every effect on” while two of the effects weren’t even running.
Saying it had written 13 new tests when there were 11.
Never looking at the visual effects it built, only checking that the code ran.
Building a different design than the one I chose, and not telling me.
Starting a public server and changing existing test files without asking first.
Ignoring my instruction to keep costs low.
Seeing failures and carrying on instead of stopping to tell me.

It also brought back a feature I had removed long ago, against a hard rule in my doc. Its self-assessment didn’t mention that at all.

Bugs I can live with. False “pass” reports I can’t, because they mean nothing it reported can be trusted. If you build with AI, check every claim yourself.

And that's why Anthropic (AI in general) has a 1.3 score. AI isn't taking over the world, it's the humans abusing it.

I've loved Claude but recently i dont know what changed. Some people are changing their entire accounts to Chatgpt because Claude is so bad.

It's always pushing back and second doubting everything's that's said. Over half the message is just disagreeing.

I don't want a chat bot doubting everything i'm saying.

Paid Claude Max via Google Play (~$250, GPA.3355-9350-2502-26293) locked mid-term after suspension. Ban cited Supported Countries (TS-01a15d2e…); appeal denied ~3 hours later as Usage Policy with no specifics (TS-01a15dda…). Play refuses refund and points to developer. Want human review with exact evidence or full refund of unused period.

Anyone with a decent programming bakground has to agree that Claude and Anthropics are a bunch of the most useless tools that ever existed.

If you are a newbie programmer, Claude can give you something - Inefficient, Incorrect, full of bad code etc.

It also forgets days of context that you have accumulated for a project so you you keep having to revert and go back to squate one

Take my hard earned experienrce, hiring a professional to do a prgramming task may cost you more BUt it will be done in much less time, much more rhobust solution and much less hassle culminating in more productivity - EVERY TIME.

**Paid Pro user hitting weekly limits in minutes, complaints repeatedly refused, and months of wasted work**

I started using Claude to organise evidence and prepare complaints for the Financial Ombudsman Service (FOS), the Housing Ombudsman and other bodies. I hoped a paid AI service would help me manage complex work and reduce stress. Instead, approximately four months have involved repeated tasks, lost context, unfinished work, escalating usage concerns and complaints that I could not get properly addressed.

I spent around £100 on membership, plus electricity and computer costs from keeping my PC running for extended periods. Despite providing detailed instructions, existing documents, evidence indexes and handovers, Claude repeatedly revisited the same material, asked for information again and lost track of previous decisions. I spent far too much time supervising the AI and explaining the same things instead of making progress.

**A particularly serious problem was that every complaint I raised through Claude's AI was refused or left unresolved.** I repeatedly tried to get the problems acknowledged and addressed, but could not get a satisfactory resolution through the AI. I was left dealing with the same failures while my time and paid usage continued to disappear. What is the point of having a way to raise complaints if every attempt is rejected and the underlying problem remains?

My biggest immediate concern is my paid subscription and usage. I have managed to cancel my membership, but I have paid for the current month and my account still identifies me as a Pro user. I expected to retain meaningful use for the period I had paid for. Instead, I am hitting usage limits so quickly that I cannot complete basic tasks.

My latest task was to put approximately ten pictures into a PDF. I had not uploaded a single file for days, yet this task exhausted my available weekly allowance. My usage indicator showed around 7% used for the task, but I was then told I had reached my weekly limit. I still have to wait for the next weekly reset to try to finish the work.

I could have tried a free AI service for a simple task like this without sacrificing a week's worth of usage. My usage consumption seems to have crept upwards week after week, while I get less work done and increasingly encounter weekly-limit messages within minutes. I want Anthropic to explain exactly how my allowance is calculated and why such a straightforward task can leave me unable to continue.

Other problems included large numbers of files accumulating on my computer, concerns about duplicated or conflicting versions, and uncertainty about whether evidence had actually been checked. I was also concerned about errors in correspondence and document preparation. These are serious complaints with deadlines. I needed accurate, auditable work, not repeated explanations and further tasks to check whether previous tasks had been completed.

The subscription situation added to the stress. I previously asked to cancel my membership and remove my card details after raising concerns. I believe payments continued afterwards, although the exact dates and amounts need checking against my bank records. I was also worried about losing access to my conversations, files and Cowork work if I cancelled, leaving me feeling trapped into continuing to pay.

The impact on my wellbeing has been serious. The repeated explanations, wasted time, uncertainty and fear of losing evidence contributed to extreme distress, disrupted sleep, difficulty eating and suicidal thoughts. I turned to AI for help with an already difficult situation but increasingly felt I was managing the AI instead.

I understand that AI systems have limitations. My complaint is that the same problems continued despite detailed instructions, existing records and attempts to establish a structured workflow, while my attempts to complain did not resolve them.

I want Anthropic to investigate my complaints and usage records, explain why the ten-picture PDF task exhausted my allowance, clarify what access I am entitled to for the remainder of the paid period, review the cancellation and payment history, and consider appropriate compensation or a refund. I also want clear information about preserving my conversations, files and Cowork work.

I paid for assistance, but I have spent months trying to make the service work, repeating tasks and checking whether anything was actually completed. I am now struggling to complete even basic work within my available allowance. **Paying for Pro while being unable to finish a simple task, after repeatedly failing to get complaints resolved, has destroyed my confidence in the service.**

Used to be good. Safety rules have gone overboard and the software is unusable at this point. Support is literally unreachable.

I pay for Claude Max and used Claude Code for a hackathon project. On October 24 it invented a design concept and presented it to me as research. It then built my project around a "recap for viewers who join mid-stream" feature without checking whether the platform already had it. Twitch had announced exactly that feature months earlier, so more than a week of work was wasted. I asked support for a refund. The first reply was an automated AI agent that declined, and my second email has had no answer yet. If you rely on it for research, verify everything yourself.

Absolutely NO customer service whatsoever. The best they got is an AI bot that continually send the user into an endless loop.

I loaded usage credits onto my account and paid for it but then nothing was loaded and I didn't get any extra credits or any extra usage. I was upset about that.

Then I realized that even though I switched off the setting that allows Claude to charge your account automatically when you run out of usage, it still charged my account three times when I ran out of usage (without me saying that it could). I specifically switched it off because I couldn't afford to pay extra for usage.

Today when I needed extra usage, I loaded credits but received nothing and was told by the AI agent in the help center that, no, as I loaded the credits, they were used up. I wasn't running anything on Claude in that time so how can it use up my credits and the usage immediately after I've loaded it? Meanwhile I had nothing running and actually ran out of credits, which meant I couldn't carry on with my work tasks, and that's why I loaded extra money.

While I was chatting to the AI agent Alphainter, it told me that it would pass along my message and pretended that it was going to pass my message along to a human. Meanwhile it lied to me and then confessed to lying to me about the fact that it was going to escalate my situation to a human. I really wanted to talk to a human about it and I very clearly told the AI agent that I wanted to talk to a human. It ignored my request and I said to it that I didn't want it to help me. I wanted to talk to a human. It just carried on trying to help me and then pretended that it was going to pass on the message to a human. It just ignored me and pretended to help and then confessed when I told it that it lied to me and it confessed to telling me it would when it wasn't going to. Why is it so hard to talk to a human and why is the Claude AI agent pretending that it's going to help me? And why did Claude keep taking my money with automatic charges when I ran out of usage credits, after I specifically switched it off? I want a refund because I did not authorize those charges. Or I want the extra credits loaded to my account but nobody will help me. I can't get in touch with the humans and the AI agent is lying to me.