It IS the official solution.
Puzzles popping up on users is not OpenAI’s problem;
bots scraping GPT-4 answers is OpenAI’s problem.
It just stinks.
Same here. I am also a plus user, and it require verification after every prompt when using GPT4. Pretty annoying.
1 Like
Yep seeing a captcha on every single query as a plus user.
Same here I will cancel my plus account at the end of this billing cycle
2 Likes
I noticed this today. Using Safari fixed the issue. I only received one Captcha at the beginning instead of after every prompt.
Testing side to side against Chrome, the problem persists on Chrome.
I’m a plus (paying) user here too.
I get prompted for human verification at every input. Right after confirming and passing the human verification, any input on chatgpt triggers another captcha verification.
I tried different browsers, incognito mode, third-party apps, the official app, etc.
I am always on the same IP address from a well-known and well-established major ISP in the country. All my last interactions have been with Opera, Chrome browser and the official smartphone app from Google Store.
Constructive criticism is welcome, but criticize ideas, not people.
Where did the idea come from? Hanging from a tree?
If I say, “Such and such is a stupid idea,” what is the reaction of the individual who had that idea? “nothing to do with the idea.”
Could it be that “we” have said something to chatgpt that deemed us not worth being kept as users, and the captcha is just a politically correct method to have users give up on chatgpt as opposed to being forcibly removed and thus making it look like “we” are not being cancelled?
How much more human verification is needed for a paying user?
Shouldn’t paying users be the ones that should not need human verification? The platform has our banking information!
This constant captcha issue is not random. It is by design and targets specific users.
This is a dangerous assumption and most likely not true.
OpenAI offloads as much as they can. So if you want to beef, it would be with Arkose Labs.
You have mentioned that it’s all under the same IP address. Out of curiosity, do you use Opera GX? I know you said Opera. I believe there’s variations though?
You can change browsers, do anything that typical bot makers do to hide their identity. They most likely do everything they can to add trackers.
Do you use extensions? Any strange conversations pop up?
1 Like
The mathematical probability of my statement not being true is less than stating that the outcome of my concluded negative experience with chatgpt is dangerous.
What evidence is there to prove that what we conclude from experience is not true?
What is really dangerous is stating that an opinion based on the individual’s experience is deemed as you deemed it since it will always lead to the shutdown of the individual who “complained” against the “established party” about his negative experience with the established party.
(It also sounds like you are criticizing me)
The next step is to label dangerous as unsafe, then to single out the individual with the “dangerous assumption” as a liar, discredit the individual and finally terminate the individual from the public eye.
(I am talking from experience on the internet as well as outside of it, and I was a power internet user before the people who created chatgpt were born)
Twitter was also full of conspiracy theories that led many individuals to be shut down. As history shows, many of those theories turned out to be true. The evidence that deemed them false is still nowhere to be seen.
There is no logical reason for openai to keep verifying a user who is verified by financial records, banking information, IP, software, browser, smartphone apps, app store account, etc.
Chatgpt is one of the most advanced tools on the internet. It is not a collection of newbie scripts, and we live in a time when open debate of ideas and discourse is not tolerated anymore.
And no, there are no extensions on any of the browsers used, not Opera GX either.
I just tried another browser without any modifications to it, and the result was the same. Also, there are no strange popups anywhere. Just captcha after captcha, which, given my eyesight problems and difficulty seeing well, gets me extra upset and sometimes failling on the last try.
I do not wish to pursue this type of conversation.
I came here with a technical problem adding my account to the list of complaining accounts and nothing else.
I have stated the problem quite well on the technical front. We who pay are more than well-verified as being human.
The collection of personal information done to paying users far surpasses the need for a captcha, and it makes no logical sense to use captchas.
Any abuse done by a paying user can be quickly resolved and better addressed than using captchas.
Here is a solution from constructive criticism:
It is rudimentary, basic and straightforward, which can be done by simple methods.
If the user’s email/registered email matches the banking account paying records on file, then bypass the captcha for the identified user.
It already works to give access to gpt4. This is how we can use gpt4 from 3.5.
Here is a purely conceptual idea using an archaic programming language (one of the first I learned) to explain the principle, and yes, I know that BASIC wouldn’t actually be used for this, as it can’t interact with databases or modern internet protocols, but the concept should be easy to understand without rhetoric. It is just a fun algorithm idea bear with me on this one.
10 INPUT "Enter registered email: "; email$
20 LET bankEmail$ = "example@email.com" REM Replace with actual email on bank records
30 IF email$ = bankEmail$ THEN GOTO 70
40 PRINT "Email does not match. Show CAPTCHA."
50 STOP
60 PRINT "Email matches. CAPTCHA bypassed."
70 STOP
It is not a new radical idea. It is used in many other platforms online. It works.
It works to prevent ads. How do we get rid of ads on YouTube? By being a paying user.
No excuse not to work on or with chatgpt with browsers or even 3rd party API apps.
Paying user = Bypass the captcha.
Can’t implement it? Don’t use a captcha.
Want to use a captcha? = Paying users will drop openAI= Loss of revenue.
Lost revenue due to ideologies = Bad business model
I am on the list to close the account if this does not get solved.
If I understand correctly, you are suggesting that OpenAI has implemented some sort of monitoring service to “judge” what you say. If you say things that “they don’t like”, they use their third-party anti-bot service by Arkose Labs to trigger an over-burdening amount of captchas so that they can kick you off in a politically correct way?
I’m all for conspiracy theories, and I agree. There are conspiracy theories that were out there that turned out to be true. That doesn’t justify anything though. This is a plain case of “whataboutism” that doesn’t apply here.
What I am saying to you is: prove it. For every deviation that you believe OpenAI is using, prove it. What I am suggesting is that they are using the service as intended. My proof is by going to Arkose Labs and learning about what they do. I am almost certain there is something strange going on in your computer.
For me, it makes complete sense, and is the least path of resistance to say that “They are using a third-party service (Arkose Labs) to catch bots and display captchas to prevent them”. Because, that’s what it does.
Of course there is. Let’s say you download some malicious software that can run headless browsers on your computer. You are a paying ChatGPT user. This software opens up ChatGPT and starts spamming conversations to max out your limit. It can sell this service for $5/month to people in other countries that don’t have access to ChatGPT. That’s one reason.
Let’s say that some people did the math and realize that hitting the limit of ChatGPT is cheaper, and more reliable than building their own API. So they build a bot to automate ChatGPT requests and cost OpenAI a lot of money.
ChatGPT is meant to be operated by a human. So it makes perfect sense why they have anti-bot security implemented. There were and probably are millions of bot attempts happening each day, costing OpenAI a lot of money.
You (an innocent person) is caught in the fray. I would like to know why. I’m sure you do as well. So let’s start with the most obvious answers, and then maybe we can work our way towards some serious conspiracy stuff instead of immediately jumping into it.
But, if you wanted to come and leave some slightly slandering open-ended questions and not want to deal with the discussions that follow, that’s fine. Whatever.
The conversation is going on a rhetorical and subjective debate, which I do not wish to go and, frankly, will not resolve the problem but rather keep us “arguing” with no solution from your side.
Paying users are identified in much better ways than using a captcha. Any abuse from those users can be rapidly taken care of in a much more effective and legal way than using captchas. (as stated before).
The argument of a captcha for user ID/verification vs a Bank account has no credible merit in any technical way or court with the rule of law.
Also as stated before, many platforms use paid subscriptions to relieve the users from unwanted features and or “annoyances.” Any abuse from those well-identified users is resolved faster and more effectively than with web browser popups. OpenAi does not want to use them? That’s a feature.
As an example to address my previous stance:
I was once asked to implement features to apply to undesirable users on a platform.
The features should ensure that the user never suspected he was being targeted. They were meant to have the user give up and blame it on the connection or software used or some service being crappy.
There was even tech support and a booklet to address these users’ problems to make it look very legit and to be used by tech support people who knew nothing about the real problem, so their human reaction was never compromised.
The help given to the target was legit but without a permanent solution. Whatever was temporarily solved would later be back in some fashion.
Eventually, the target would give up and go away. Clean as a whistle.
This was one of my experiences elsewhere, and I really do not want to be fueling this type of conversation here. Leads to no solution.
Here is another way to resolve the problem on the rhetorical front.
Usually, this way motivates people to move faster to resolve problems.
All the paying users, companies, and businesses being illegitimately targeted with captchas in a fashion that makes the service unreasonably useable and or completely unusable should get together and launch a class action lawsuit against openAI based on any of the following:
* False service advertisement vs outcome * Loss of productivity during work time * Discrimination towards people with disabilities (vision, hearing, etc) <-- This one is the best. Legally, anything can be used these days. * Paying for a service that fails more often than it works * Loss of company revenue due to failing services provided etc etc, the list of ideas is long.I know it will all be debatable by the masters of rhetoric in court. But they are paid to do that, and if we hire the right psychopaths, it is money in the bank!
It will also cost money to openAI to address the case and much more than resolving the captcha problem. (No to mention bad publicity)
Then we would see how much longer we would have any captcha with chatgpt.
I don’t know what else to say to resolve the problem that I have not said before.
Sure, we want it resolved. But how?
Same user, same IP, same OS, same email, same bank account, same browsers, no add-ons, all standard access, same software, same hardware.
There is nothing I am doing that could trigger a constant captcha response and, even worse, trigger one captcha after another on an endless loop.
I get fewer captcha using Tor and Tor browser to access mainstream platforms such as but not limited to Gmail, YouTube, Facebook, etc.
Google is not worried. Once the user is verified even using Tor, life moves on.
This problem is a feature. It will disappear when the feature is removed, or legit paying accounts can bypass captchas just like they bypass gtp 3.5 to access gpt4.
There is nothing else to say about this unless the intention is to make me give up on the rhetorical front or entrap me in the community guidelines to ban me legitimately, thus proving one of my points.
The feature is both a failure and a design. Removing the captcha for paying accounts is a basic solution.
Accusing, suspecting and or suggesting that paying accounts will abuse the platform without any or substantial evidence and, therefore, justifying the use of captchas is a wrong route to take the business onto. Not everyone who logs in is a criminal.
I have provided two constructive critical solutions for a problem caused by an implemented feature by design and the desire of openAI to use it.
One is technical and the other one is legal.
The third option has been stated before here as well as in many other places online, which is to cancel the subscription.
Look. I get that you’re upset. I am not trying to ask for your opinion or how you feel. I am asking about your browser and computer environment.
It was never implied that everyone who logs in is a criminal. Nobody is accusing anyone here of anything. Do you have locks on your doors? Basic security is essential. It does not imply anything else.
For the third time, the captcha you are hitting (you and maybe 0.05% of users) is typically for catching malicious entities (bots, mostly). It has nothing to do with being a paying user or identification.
This system most likely blocks out millions of malicious requests, and unfortunately you are caught in it. The majority of users do not have have any captcha issues.
You are an anomaly. If you could just be a little more descriptive about your computer maybe other people encountering the same issue with say “Hey, I have the same thing!” and we can actually solve this problem.
I have a sense that tells me that you like to tinker with stuff. So do I. The only times I have been bombarded with captcha requests was when I was tinkering. Stuff as small as switching to responsive mode while using ChatGPT. My suspicion is that you have tinkered with something which raised a flag.
This is fairly descriptive of the problem at hand. It’s a lot more than what all other users have said, such as (paraphrasing here) “it is not working for me… etc etc”.
And I am not upset. Annoyed yes.
I was also very descriptive about the browsers used and in what conditions and answered your questions.
For the record, I have addressed this issue more technically than everyone else here stated until I came here.
I don’t think doing deep packet inspection/traffic analysis on my end will contribute to a better report, nor will I do it.
The problem is not on our side. Neo was also an anomaly.
I have reported the problem and provided two solutions, one being technical and reasonable and widely implemented on the internet by the majority—of well-established platforms.
This interaction has not contributed with any value to a solution to my/our problem.
Any problem in life is only sustained if all parties involved decide to feed it. There is a button on our end that ends the food supply.
I hope you have a great rest of the day or night, can enjoy quality time with what you enjoy more in life and have a great rest of the week. 
They already know that you are paying. Bad actors are also the ones paying (or stealing your account credentials).
OpenAI also knows that for $20 a month you get 400 GPT answers a day, 12000 GPT answers a month if you run at your limit 24/7, something that would cost $2000 at the size of the inputs that account hackers run on stolen API accounts that have been shown here.
They cannot allow bots to train Chinese AIs using ChatGPT.
1 Like
This started with me recently too, I think it’s because OpenAI detects “unusual activity”. In my case it ONLY BEGAN when I used chatGPT on cellular data last week while moving in a car. Ever since then every prompt on GPT3.5 has asked me to “verify that I’m human”
They really think I’m a bot because of a ‘moving IP’ 
Clearly they haven’t accounted for mobile hotspots. Interestingly enough, this “verify that I’m human” prompt only occurs on GPT3.5 for me and not GPT4… I suspect they use different servers for each, and the 3.5 server detected me as ‘suspicous’ but not the 4 server. Funny
This human captcha thing is driving me crazy! Every question I asked it stopped to verify I was a Human. And I used it a lot and loved it. Now I wouldn’t say I like ChatGPT as a result. Really annoying!! This will switch me to another product
I have the same frustration, and I cannot get an answer from anyone at OpenAI.
Does anyone know how to reach someone? All i get is bulls–t boilerplate. I am using Bard for a lot of things instead of ChatGPT.
I’ve read that a busy system can account for the verification demand, but we pay $20 a month to enhance the experience.
OpenAI is giving AI platforms a bad name.
same here. thats a bad joke since i am paying for the plus plan.
Same here. But it does not ask to verify while I am using it in my company network. It asks me only when I am using it at home.
Don’t even bother contacting OpenAI for this, all they ever do is paste the same copypasta message every time with useless troubleshooting (clear cache, try different browser, etc.) I get no captchas anywhere, only on ChatGPT and only when I try to access the premium features.
It feels like a spit in the face, “thanks for the money, but we won’t allow you to use what you paid for!”
I contacted billing once more and make a screenshot with timestamps every time I am locked out of the service and will pass this onto my bank shall they not fix the issue.
This forum is comprised of fellow users.
For account issues, like demanding a refund due to CAPTCHA, you would go to help.openai.com and use the assistant icon at lower right to send a message to support staff.
1 Like