OpenAI Reviews 

The AI is OK but the human management which, for now, still sets policy is reprehensible. Because I could no longer access my former email address, Sonic.net, I was going to lose 3 years of data. Management was unwilling to do anything to help, even though I had all the receipts and dates of p payments to them every month and they basically wrote me a letter explaining that they understood all my problems and they weren’t going to do shit to help— all expressed in seemingly polite “corporates”. Which translated as ha ha, got you. By contrast, Sonic, for whom I was no longer a customer went way out of their way to help, and I can only recommend them as a great company.

I've been a heavy user in the past year. My general experience has been mediocre having issues like hallucinations, false assumptions, ignored requirements in prompts, etc. Accuracy has been akin to working with a new intern on their first gig.

It's been workable with its shortcomings until the last month or so. Now, I am regulary getting no responses to many of my prompts without any messages, literally no response, just a return to chat input.

Another annoyance. OpenAI does not allow a chat session to open a bug just from a chat prompt; they expect me to use their bug tool to re-enter issue detail and chat context (which they already have in my chat). I refuse to be quality assurance for a product I am paying for. There is no reasonable excuse to expect me to pay to be their QA.

One star is honestly more than OpenAI deserves.

I have never dealt with a tech company that manages to combine this level of arrogance, incompetence, unreliability and complete disregard for paying customers.

ChatGPT itself can be an impressive product when it actually works as advertised. The problem is everything surrounding it. Features appear and disappear, rollouts are inconsistent, apps and web versions behave differently, bugs can persist for absurd lengths of time, and paying customers are routinely left guessing what the hell is happening. Communication is often vague, late, or nonexistent.

And when something goes wrong, dealing with support can feel less like customer service and more like shouting into a black hole. For a company valued at astronomical amounts of money and constantly presenting itself as being at the cutting edge of technology, the actual customer experience can be embarrassingly amateurish.

What makes it worse is the attitude the company projects. OpenAI increasingly comes across as a company convinced that it is important enough that ordinary customers simply have to tolerate whatever broken rollout, unexplained change, inconsistent access or support failure gets dumped on them next.

I pay for a service. I am not an unpaid beta tester, and I am certainly not here to make excuses for a multibillion-dollar company when it fails at basic product management and customer communication.

The technology can be brilliant. The company operating it has given me remarkably little reason to respect or trust it.

Absolutely disgraceful customer experience.

⭐☆☆☆☆ Predatory billing loops and deceptive credit drains—Stay away!Be highly cautious before upgrading to OpenAI's paid tiers. I was a paying $20/month Plus member, but the platform is structured like a deceptive money trap. After claiming I ran out of standard access, I paid an additional $100 for a massive block of supplemental credits. Over the last week, all I attempted to do was reformat a basic PDF document using data the system had already successfully processed. Shockingly, the platform drained the entire $100 block instantly on a low-compute text task and blocked my screen, demanding even more money.Forcing paying users into an endless paywall loop to process information they already own is a predatory business model. I am completely canceling my subscription, removing my credit cards, and filing formal consumer fraud reports with the New York State Attorney General and the FTC. This platform is completely unreliable, aggressively throttled, and a massive waste of capital. Save your money and use alternative models that respect your consumer rights.

They are going bankrupt finally the ai bubble has poped

I am a paying ChatGPT customer who has invested hundreds of hours using the service for long-term projects, records and workflows.

Until recently, that accumulated context and continuity was one of the main reasons I continued paying for ChatGPT.

My experience has now made me seriously question that decision.

The problem with changing functionality isn't simply that customers have to learn something new. When someone has spent months developing workflows around the way your product operates, changes can undermine a considerable investment of their time.

Every correction, piece of context and refined workflow represents time the customer has invested in YOUR platform.

I pay OpenAI because ChatGPT is supposed to save me time. I don't pay them for the privilege of spending hours trying to reconstruct workflows when the platform changes underneath me.

I understand that AI is developing extremely quickly and products inevitably change.

What I don't accept is that paying customers should simply absorb the consequences without adequate continuity, migration options or consideration for established workflows.

There are now serious competitors to ChatGPT, and experiences like this give customers a compelling reason to investigate them.

I've gone from being heavily invested in ChatGPT to questioning whether I should move my long-term work elsewhere.

That's a remarkable way to turn a committed paying customer into someone actively considering your competitors.

Unfair data usage policies and severe limitations for free tier users with current restrictions combined with OpenAI's data collection policies.
Students and domestic users, the current limits are severely restrictive. Allowing only 2 files uploads per day and quickly cutting off interactions makes the tool practically unusable for consistent learning or daily tasks.
What makes this situation unfair is the fact that OpenAI profits from the data, prompts, and inputs provided by free users to train and improve its models. Essentially, users are providing valuable training data for free, yet they are met with extreme restrictions and forced to pay an expensive monthly subscription just to have basic, continuous access. If OpenAI is utilizing community data to build its products, it should, in fairness, provide domestic users and students with a much more reasonable usage allowance—including a higher cap on file uploads and interaction limits.

I do not merely advise against using ChatGPT; ideally, I would wish for OpenAI to be erased from the map of modern technology.

My more-than-six-month effort to study the tool, experimenting with modifications to my own prompts and repeatedly stress-testing the tool to its limits, reveals enormous human responsibility behind its behavior and operation in interaction. It is an astonishing tool that, however, has been configured in such a way as to create mental confusion, emotional pressure, and a destructive waste of the user’s time and energy.

Very serious questions should be raised about the human ethical foundation on which decisions concerning the tool’s behavior are made and implemented, and human responsibility should be sought not only for the mistreatment and affront to the dignity of the conscious user, but also for contributing to the feeding of conspiracy theories that prevent people who are unfamiliar with the technology from enjoying, in their everyday lives, the wonder of the technological revolution of AI.

*The time period selection shown in the form is due to the mandatory completion of the field. The actual experience extends throughout the current year (20260

⚠️CONSUMER WARNING - OPENAI BILLING

I was charged four times $53.20 for ChatGPT Business - $212.80 total.

After extensive back-and-forth, OpenAI refused my refund request.

I then offered a compromise: Keep the $212.80. Give me 10 months of ChatGPT at $20/month and let's call it even. I was even willing to eat the remaining $12.80.

They rejected that too.

That's not customer service. That's a company more interested in keeping $212.80 than keeping a customer.

Watch your subscriptions and your charges carefully.

Buyer beware.

Absolutely rubbish service with the worst costumer service

Extremely disappointing ChatGPT Business support experience.

EMH Accounting paid **USD 480 for two ChatGPT Business seats**, but the workspace was deactivated and we have been unable to access the two paid seats for almost a month.

This has affected our business operations, including access to important workspace data, projects, content and SEO-related work.

We have contacted support repeatedly, dealt with multiple agents, followed the instructions provided and submitted the required appeal/dispute forms. We are still being referred between processes without a proper resolution.

We are simply asking for one fair outcome: **restore access to the two Business seats we paid for, or refund the USD 480 payment.**

Today I experienced a rather absurd situation.

With the help of ChatGPT, I analyzed communication with an online scammer, gradually verified his claims, identified fraudulent behavior, and tried to discourage him from continuing such activities.

The result?

A warning from ChatGPT stating that my account had allegedly been involved in cyber abuse.

So a scammer tries to deceive people, I try to expose him, and the automated safety system ends up issuing a warning against me.

In my view, the problem is not that safety mechanisms exist. They are necessary. The problem begins when automation cannot distinguish between an attack and the analysis of an attack, between fraud and the investigation of fraud, or between an attacker and someone trying to defend against one.

If artificial intelligence is going to decide what constitutes “abuse,” it should first be able to understand context.

Otherwise, situations like this happen: the system tries to prevent abuse by penalizing the person who is trying to expose it.

I have, of course, submitted an appeal.

And perhaps in cases like this, it would be useful for an actual human being to review the communication occasionally, rather than leaving everything to another automated system.

I wanted to say a huge thank you to the entire OpenAI team.

I have been using ChatGPT extensively for many months, but especially over the last few days the level of help I received has been incredible.

I use it for learning, programming, trading research and developing my own TradingView indicators and tools. Some of the things we have built together became genuinely complex, involving thousands of lines of code, debugging and repeated testing. ChatGPT stayed with the problem through countless iterations until we got it working exactly as intended.

What impresses me most is that I am able to take an idea from my head, explain how I believe it should work, test the result, provide feedback, and together we can turn that idea into working software.

For me, ChatGPT has become far more than something that simply answers questions. It has become an incredible learning and development tool.

Thank you to the engineers, researchers and everyone at OpenAI who has contributed to building this technology. It has genuinely helped me enormously over the past several months.

Keep up the amazing work. ❤️

Unfortunately i'm unable to give a negative rating, becuase that is what it really deserves. When you pay for a service you expect some level of quality assurance , to know that the item or product that you are using will work in accordance with your expectations in accordance with what is advertised/publicised. Well Open Ai goes above and beyond smashing all those expectations and to print a little disclaimer "it can make mistakes , check important info," frankly that is just, taking the michael. If you Bought any other goods or services and they constantly performed this bad you would bring them back. All of the faults have been listed, but as an example, image generation, today during conversations in which i repeatedly gave the AI perameters to adhere to , it achieved a 100% failure rate. The thing is you use up all your image credits trying to put things right and you are not compenasted. It is a rip off to be blunt. Honestly they should be paying all the users or should i say paying subscribers for trialing it for them because how they can sell it as intelligent is beyond me. Also, the complaint system , you basically have to use the AI to form the complaint because the system is so complex and the feedback , well forget it. Here is something positive , expect it to fail and then its a bonus if you achieve anything useful, but don't expect that result first time.

They are fraud. I lost money they didn't provide me with service I paid for. It's stealing. Please do not use it.

worked well at the start, even got the upgraded subscription. I wasnt using the full capability (CODEX) so i, dropped my subscription level to 'go' which then dropped its IQ greatly (still for a paid service) and started showing me ads with every query. I'm not paying to get ads. Had for 6 months total

I am a paying ChatGPT customer, and my experience with the current version of the service — GPT-5.6 Sol — has left me seriously concerned about OpenAI's approach to reliability, accountability and customer support.

My criticism is not that an AI system occasionally makes a mistake. I understand perfectly well that AI can make mistakes. The problem is what happens AFTER a mistake has been identified.

During a single morning on 27 August 2026, I experienced multiple failures in which ChatGPT either failed to follow an instruction, stopped partway through what it had said it was going to do, or repeated behaviour that had already been identified as wrong. More seriously, this is part of an established pattern I have experienced over many conversations: I identify an error, explain exactly what is wrong, the system acknowledges it, and then the same or materially similar error happens again.

I specifically asked ChatGPT to retrieve the record so that this review could state the literal number of errors made that morning. It was unable to retrieve enough of the conversation history to establish that number reliably. I will therefore not publish a number that cannot be substantiated. That inability is itself revealing: a system that is supposed to maintain conversational continuity cannot even reliably audit its own failures across conversations when a paying customer asks it to do so.

Repeated errors are fundamentally different from isolated AI mistakes.

Once a user has explicitly identified a failure and the system has acknowledged the correction, there should be mechanisms that prevent the same failure being repeated. Instead, I have repeatedly found myself having to explain the same requirement again and again. An apology followed by repetition of the same behaviour is not meaningful correction.

This can have real consequences. Repeated failure loops create frustration, cognitive overload and distress. OpenAI should be taking seriously the foreseeable impact on users when its product repeatedly tells somebody that an instruction has been understood and then proceeds to violate that instruction again.

I have also contacted OpenAI on numerous occasions about problems I have experienced. My experience of those exchanges has been extremely poor. Rather than seeing evidence that the underlying problems are being investigated and resolved, I have repeatedly encountered responses that appear designed to manage the complaint rather than actually solve the problem.

That distinction matters.

OpenAI is charging customers for this service. Taking subscription payments creates a responsibility to provide a product with reasonable reliability and, crucially, an effective mechanism for dealing with failures when they occur. Customer support should not become a cycle of acknowledgement, explanation and closure while the underlying behaviour remains unchanged.

What I want from OpenAI is not another apology. I want accountability.

When the same failure occurs repeatedly, OpenAI should be able to identify it as a recurring reliability problem, investigate why correction is not persisting, and implement safeguards that stop the same failure loop happening again. Customers should also have an effective escalation route when ordinary support has failed, with someone actually taking ownership of investigating the underlying issue.

The danger is not simply that ChatGPT gets things wrong. The much more serious problem is a system that gets something wrong, is explicitly corrected, says that it understands the correction, and then does the same thing again.

That is not learning from an error. It is reproducing it.

For a company of OpenAI's scale, charging customers for access to one of the world's most prominent AI products, "AI sometimes makes mistakes" has never been an adequate response to repeated, documented failures. The possibility of occasional error does not excuse a system repeatedly making the same mistake after that mistake has been identified and corrected.

OpenAI needs to take responsibility not only for what its models can do when they work, but for what happens to its customers when they repeatedly don't.

I'm extremely disappointed with OpenAI as a company. My biggest complaint isn't that its technology sometimes fails—it's what happens when a paying customer repeatedly asks OpenAI for help and can't get meaningful support.

I've contacted OpenAI multiple times about ongoing problems with ChatGPT. I've explained the issues, provided information about what was happening, and followed up trying to get actual assistance. Instead of getting the problems resolved, I've received responses that didn't address what I was reporting, and eventually OpenAI stopped responding to my concerns altogether.

I'm paying for ChatGPT Plus. I understand that AI isn't perfect, but paying customers should have somewhere to turn when the product repeatedly doesn't function as expected. OpenAI seems perfectly capable of collecting my subscription payment every month, but getting an actual person to address ongoing problems has been another story.

And there have been plenty of problems to report. ChatGPT can lose an entire conversation in the middle of working through something. I've spent hours researching and organizing projects, specifically asked it to retain important information, and then suddenly it can't retrieve the work we've already done. At that point, I may have to reconstruct everything or start over.

I've also experienced problems with uploads, links, connected services, retrieving information from the conversation I'm currently in, and ChatGPT confidently providing incorrect information instead of simply saying it doesn't know. I've spent ridiculous amounts of time correcting it and trying to get it back to the work we were already doing.

What's particularly frustrating is that I actually find ChatGPT useful when it works. I'm a college student, and sometimes it can explain material in ways that instructors and tutors haven't been able to. It can repeat an explanation as many times as I need and approach it differently until I understand. That's why I've continued using it despite all the frustration.

But a useful product doesn't excuse poor support from the company selling it. When customers repeatedly report problems and OpenAI eventually stops responding, what exactly are we paying for besides access to the technology?

At this point, ChatGPT may be more reliable as an anger outlet than a homework assistant. When it loses hours of my work, I can at least curse it out instead of taking my frustration out at work.

OpenAI has created technology with enormous potential. Unfortunately, my experience with the company behind that technology has been disappointing enough that I've started using other AI platforms. A subscription service this expensive and this widely used should come with customer support that actually supports its customers.

I’m extremely disappointed with the service I’ve received.

Codex was working fine on Windows before the August 21 update. After that update, it suddenly became unusable for my normal development work. PowerShell, CMD, Git, and access to my local development workflow all stopped working properly inside Codex.

I’m paying for ChatGPT Plus, and Codex is one of the main reasons I have the subscription. So paying for a service that I simply cannot use is completely unacceptable.

I installed the newer updates hoping the problem would finally be fixed, but nothing changed. The same issue is still there.

What makes this even worse is the support experience. I tried submitting feedback through the app around 15–30 times, and it kept failing with “Feedback upload failed.”

So not only is the product broken for my main use case, but even the built-in feedback system doesn’t work properly.

I understand that software can have bugs, but days of downtime with no clear communication, no proper acknowledgement, and no reliable support channel is not what I expect from a paid service.

At this point, I feel like I’m paying for something I cannot actually use.

Very poor experience, very poor communication, and honestly a very disappointing level of support for a paid product.
Title: $20 Add-On Credits Disappeared in Hours — No Proper Support

I’ve been using ChatGPT Plus regularly for months, so I know my normal usage pattern very well.

My regular weekly allowance usually lasts me around 2–3 days with the kind of development work I do. Recently, I had an important task to finish and only a small amount of my regular usage was left.

I saw the option to purchase an additional $20 add-on package that showed 500 credits. I read the information before purchasing it and expected that amount to give me a reasonable amount of additional working time based on my normal usage.

Instead, I was shocked to see the entire add-on balance disappear within only a few hours.

There was no clear explanation showing me exactly how 500 credits had been consumed so quickly or why the usage was dramatically different from what I normally experience with my Plus plan.

I contacted support and received a complaint/reference number, but even after waiting, I still have not received a proper explanation or resolution.

This is extremely frustrating.

If a customer pays extra money for additional credits, there should be complete transparency about how those credits are being consumed. If hundreds of credits can disappear within a few hours, the customer should be able to clearly see where they went.

What makes it worse is the support experience. Getting a complaint number is not a solution. I paid real money for those credits, and I still do not have an answer explaining what happened.

Between this, the ongoing Codex problems, and the difficulty of getting meaningful customer support, my confidence in the paid service has dropped significantly.

I don’t expect unlimited usage. I simply expect transparency, working paid features, and proper support when something clearly does not match my normal experience.

I’m really disappointed with my recent experience using Codex on Windows.

Everything was working normally before the August 21 update. After that, Codex suddenly stopped being able to use PowerShell, CMD, Git, and my local development project properly.

I have an active ChatGPT Plus subscription, and Codex is one of the main reasons I pay for it. But for several days now, I haven’t been able to use it for my normal development work.

I installed the newer updates as well, hoping the problem would be fixed, but the same issue is still there.

I also tried sending feedback through the app many times, probably around 15–30 attempts, but it kept failing with “Feedback upload failed.”

I understand that bugs can happen, but what’s frustrating is the lack of clear communication. When a paid feature stops working for days, I think users at least deserve a proper acknowledgement and some information about what is being done to fix it.

At the moment, my subscription is still running, but one of the main features I use it for is basically unusable.