Ask HN: ChatGPT Plus now showing captchas constantly?
I noticed a recent development and wanted to ask whether anyone else has a similar experience. I have been a paying ChatGPT Plus subscriber next to some API use since they have offered this product and have never had any issues, never used the service for spamming or in what I'd consider a manner incompatible with their TOS, and have in the roughly six months rarely hit the four-hour usage limit, certainly not in the last two weeks. Additionally, I am not currently connected via a VPN or a corporate internet connection
Despite all this, I have now noticed that any request that goes beyond a quick text answer requires solving a captcha [0]. Additionally, even after solving one and getting output (that has become less reliable in my personal test on top), I get another one before any additional request gets processed. Previously, OpenAI has enforced a Cloudflare captcha for accessing the page, either during high traffic times or when accessing the site via a public VPN, but never directly after a request.
I wanted to ask whether any of you have experienced something similar. For me, at least, captchas are a massive nuisance. They break the flow I like to use LLMs for (back and forth bouncing of ideas), are an accessibility nightmare, and are generally barely acceptable for free websites that need to limit fraudulent traffic. On a paid service with a reliable, long-paying customer, it is frankly insulting and makes switching fully to Anthropic more attractive than ever, even though gpt-4o still does certain tasks in my field more reliably with less prompting compared to Claude 3.5 Sonnet or 3 Opus.
Am I the only one affected, or are you encountering something similar lately?
[0] https://imgur.com/S6qmCYy
This has been happening for a long time. I once wrote a bypass for scraping: https://github.com/acheong08/funcaptcha
A still working fork: https://github.com/xqdoo00o/funcaptcha
I do apologize for possibly playing a part in causing OpenAI to put up such annoying captchas even after using Cloudflare due to large scale automation of account creation & usage before an API was released…
I've had exactly the same experience today, having been asked to complete a captcha (of a type I haven't seen before—rotating a 3D object) after every prompt. I'm logged in and a paying subscriber of more than a year. I assume it is a bug.
I've solved multiple thousands of OpenAI captchas going back for months. It is not a new thing for me.
Paid support never responded.
Last time this happened to me, my system clock was wrong.
Thanks for the tip, doesn't seem to be the case currently, but I'll keep an eye out in that area.
ChatGPT has been painful, especially at login, for me for months. Had to move to alternatives as a result and cancel my sub. I want to come back though!
Ah the irony of defending an AI with a method that an AI should be able to solve. So the captcha solver will own the LLM, it's evolution baby!
It's more likely that they're using the free tier, and your time in Captcha is helping them train a model and prevent misuse (e.g. shared accounts, scriptures)
Leaning towards the former more.
Sadly not, have been paying for ChatGPT Plus since it's become available in my country. If I were merely a free customer, I'd have full understanding for such meassures.
Oh boy a new level of ouroborous enshittification!
LLM websites need more profit and more data, so all they have to do throw in new training questions to force their already paying customers to do more work for them, before they can do their own work! I can already feel microsofts drool on my neck
I think they're just farming human data.
Honestly? I think that with OpenAI's track record we should assume that whatever the worst, scummiest answer is is probably the truth.
It's not even that paranoid. Google has literally been doing it for a while. It's not like they gave us all those photos of traffic lights because bots can't do it.
OpenAI farmed pretty much all the quality text out there for GPT-4. They did GPT-4o built natively for multimodal, likely because they ran out of text and need to farm videos & audio. Captcha data is pretty valuable too if only to beat captchas.
There are paid services for bypassing captchas (usually for the purposes of scraping, etc).
I wish Mozilla or someone would securely integrate those into a browser. It’s only a few cents per captcha, and I’d happily pay a few dollars a month to never see a captcha again.
The browser producer could profit share. This seems way less sketchy to me than acquiring a user-tracking advertising company like Mozilla just did.
I have the same experience. I never saw a captcha in the past couple months but in the last couple of days I'm getting a lot of them.
Could this be a browser issue? Are you using Firefox or have certain kinds of tracking disabled, perhaps? I think the Cloudflare captcha is especially annoying for some of those configurations.
I use the ChatGpt desktop app (on macos) many times a day and also the Android app, and have never seen a captcha from them at all. In the US, if that matters.
Yes. I've seen two today. Never before. Paid user.
I'm hoping it's a configuration error on their side. Otherwise I will definitely switch at least the client. Paying and getting captcha is a complete non starter!
Does it even make sense to scrape ChatGPT as a paying customer? From what I've heard, you would get a much more economical result (and greater programmability) by using the API.
I use ChatGPT plus pretty much daily and I haven’t been promoted for a captcha regularly. I even use multiple devices (never concurrently) and am not promoted for a captcha response.
I got 3 captcha in a row in the same chat session earlier this morning. I'm a paying customer using ChatGPT Plus. In the afternoon it look it was solved.
I canceled my subscription because of this, I couldn't be bothered TBH.
Yeah more and more