Reported a flaw to OpenAI that lets users peek at others' chat responses. Got an auto-reply on May 29th, radio silence since. Issue remains unpatched :(
Avoided their bug bounty due to permanent NDAs preventing disclosure even after fixes. Following standard 45-day disclosure window—users should avoid sharing sensitive data until this is resolved.
jonrouach · 4h ago
you're sure it's not their "feature" that calling the api with empty string returns random hallucinations?
No, definitely not the empty string hallucination bug. These are clearly real user conversations. They start like proper replies to requests, sometimes reference the original question, and appear in different languages.
jonrouach · 3h ago
i had the exact same behavior back in 2023, it seemed like clearly leakage of user conversations - but it was just a bug with api calls in the software i was using.
There was an issue with conversation leakage, though. It involved some bug with Redis.
I felt like it was a huge deal at the time but it’s surprisingly hard to quickly google it.
Sebguer · 1h ago
It was the classic "oh no we did caching wrong" bug that many startups bump into. It didn't expose actual conversations though, only their titles: https://openai.com/index/march-20-chatgpt-outage/
postalcoder · 20m ago
ah there it is. thanks for jogging my memory. funny to think of how niche chatgpt was considered then to now.
JyB · 4h ago
I don’t see anything here that would prevent a LLM from generating these. Right?
requilence · 3h ago
In one of the responses, it provided the financial analysis of a not well-known company with a non-Latin name located in a small country. I found this company; it is real and numbers in the response are real. When I asked my ChatGPT to provide a financial report for this company without using web tools, it responded: `Unfortunately, I don’t have specific financial statements for “xxx” for 2021 and 2022 in my training data, and since you’ve asked not to use web search, I can’t pull them live.`.
Sebguer · 3h ago
Do you understand what a hallucination is?
jojobas · 2h ago
Coming up with accurate financial data that you can't get it to report outright doesn't seem like one.
Sebguer · 1h ago
Models do not possess awareness of their training data. Also you are taking at face value that it is "accurate".
refulgentis · 2h ago
I don't understand the wording
Accurate financial data?
How do we know?
What does using not-web-search not having the data have to do with the claim that private chats with the data are being leaked?
01HNNWZ0MV43FF · 57m ago
> I found this company; it is real and numbers in the response are real.
???
refulgentis · 7m ago
Which of my questions does that answer?
999900000999 · 4h ago
Users should always avoid sharing sensitive data.
A lot of AI products straight up have plan text logs available for everyone at the company to view.
ameliaquining · 4h ago
Which ones? Do you just mean tiny startups and side projects and the like or is this a problem that major model providers have?
pyman · 4h ago
It's not just about sensitive data like passwords, contracts, or IP. It's also about the personal conversations people have with ChatGPT. Some are depressed, some are dealing with bullying, others are trying to figure out how to come out to their parents. For them, this isn't just sensitive, it's life-changing if it gets leaked. It's like Meta leaking their WhatsApp messages.
I really hope they fix this bug and start taking security more seriously. Trust is everything.
milkshakes · 2h ago
maybe you should stop trusting random people on the internet making extraordinary claims without proof then?
This is going to be subject to the legal discovery process with the usual safeguards to prevent leaks; in particular, the judge will directly supervise the decision of who needs access to these logs, and if someone discloses information derived from them for an improper purpose, there's a very good chance they'll go to jail for contempt of court, which is much more stringent than you can usually expect for data privacy. You can still quite reasonably be against it, but you cannot reasonably call it "plain text logs available for everyone at the company to view".
baby_souffle · 2h ago
Isn't "assume vulnerable" The only prudent thing to do here?
milkshakes · 2h ago
everything is vulnerable. the question is, has this researcher demonstrated that they have discovered and successfully exploited such a vulnerability. what exactly in this post makes you believe that this is the case?
refulgentis · 2h ago
No? Yes? Mu?
After some hemming and hawing, my most cromulent thought is, having good security posture isn't synonymous with accepting every claim you get from the firehose
fcpguru · 5h ago
well done, sounds very reasonable and following the rules.
requilence · 4h ago
Appreciate it. Just trying to do the right thing by both OpenAI and users here.
com2kid · 1h ago
I see other users conversations on my Gemini dashboard, not sure who to even complain to.
Software quality is... Minimal now days.
poniko · 4h ago
The NDA part feels really murky.
tptacek · 4h ago
It's pretty standard for bounty programs. If you don't like it, which is reasonable, do what this researcher did and just post independently.
asadotzler · 3h ago
That's an exaggeration. Most industry leaders do not require NDAs, only coordinated disclosure.
Mozilla's program, which has been around longer than most, doesn't. Google and Microsoft don't. Meta and Apple don't.
This is water carrying, intentional or not, for a terrible practice that should be shamed, so that it doesn't become standard.
tptacek · 3h ago
My understanding is that all Bugcrowd bounties do by default.
You can shame it all you want, but you can also just publish your bugs directly. Nobody has to use the Bugcrowd platform. You don't even have to wait 45 days; I don't buy these "CERT/CC" rules.
pyman · 4h ago
The bug bounty world is a funny one. I remember one complaining that their bug was dismissed and fixed after they signed an NDA, no payout, nothing. Another one got $100 instead of $5,000 because the company downgraded the severity from high to low. So they ended up with little or no money, and no recognition either. Not sure if these were edge cases, but it does make you wonder how fair the process really is.
tptacek · 3h ago
If you're dealing with large companies, a good rule of thumb is that the bounty program is incentivized to pay you out. Their internal metrics improve the more they pay; the point is to turn up interesting bugs, and the figure of merit for that is "how much did we have to spend". At a large company, a bounty that isn't paying anything out is a failure.
All bets are off with small random startups that do bug bounties because they think they're supposed to (most companies should not run bounties). But that's not OpenAI. Dave Aitel works at OpenAI. They're not trying to stiff you.
Simultaneous discovery (either with other researchers or, even more often, with internal assessments) is super common. What's more, you're not going to get any corroboration or context for them (sets up a crazy bad incentive with bounty seekers, who litigate bounty results endlessly). When you get a weird and unfair-seeming response to a bounty from a big tech company, for the sake of your own sanity (and because you'll probably be right), just assume someone internal found the bug before you did, and you reported it in the (sometimes long) window during which they were fixing it.
pyman · 3h ago
Interesting insights, thanks for sharing
maxlin · 4h ago
Permanent NDA's? Oof. It's like their plan is to just try to force the lid down till they reach ASI or something lol
tptacek · 4h ago
Again: NDAs are bog standard bounty terms.
thorum · 4h ago
> The leaked responses show clear signs of being real conversations: they start with contextually appropriate replies, sometimes reference the original user question, appear in various languages, and maintain coherent conversational flow. This pattern is inconsistent with random model hallucinations but matches exactly what you'd expect from misdirected user sessions.
A model like GPT-4o can hallucinated responses that are indistinguishable from real user interactions. This is easy to confirm for yourself: just ask it to make one up.
I’m certainly willing to believe OpenAI leaks real user messages, but this is not proof of that claim.
requilence · 3h ago
In one of the responses, it provided the financial analysis of a not well-known company with a non-Latin name located in a small country. I found this company; it is real and numbers in the response are real. When I asked my ChatGPT to provide a financial report for this company without using web tools, it responded: `Unfortunately, I don’t have specific financial statements for “xxx” for 2021 and 2022 in my training data, and since you’ve asked not to use web search, I can’t pull them live.`.
krainboltgreene · 12m ago
I’m struggling to understand why you are so adamant that this is proof.
robertclaus · 4h ago
Ya, hard to know how to react without more information.
astrange · 2h ago
GPT-4o's writing style is so specific that I find it hard to believe it could fake a user query.
You can spot anyone using AI writing a mile away. It stopped saying "delve" but started saying stuff like "It's not X–it's Y" and "check out the vibes (string of wacky emoji)" constantly.
wavemode · 1h ago
LLMs are trained and fine-tuned on real conversations, so resembling a real conversation doesn't really rule out hallucination.
If the story in OP about getting a company's private financial data is true (i.e. the numbers are correct and nonpublic) that could be a smoking gun.
Either way it's a bad look for OpenAI to have not responded to this. Even if the resolution turns out to be that these are just hallucinations, it should've been investigated and responded to by now if OpenAI actually care about security.
pyman · 4h ago
> A single misconfiguration can leak thousands of sensitive conversations in seconds. Treating privacy as an afterthought is untenable when the blast radius is this large.
Massive security bug, well spotted. It's like Bank of America showing other people my transactions, or Meta leaking my WhatsApp messages.
This raises some serious questions about security.
ajdude · 3h ago
> I am issuing this limited, non‑technical disclosure:
> No exploit code, proof‑of‑concept, or reproduction steps are included here.
Then why bother? I feel a bit cynical here, but if the goal is to get this fixed, they're not going to care unless it becomes a zero day and is given to the masses, otherwise it's going to quietly be exploitable by the few unsavory groups who know of it and will never be patched. Isn't the whole point of responsible disclosures to give them a time clock to get this situated before actual publication? Forgive me if I'm wrong, I haven't been in that field in a long time.
tptacek · 3h ago
This is the security equivalent of getting Google support by getting something to the top of HN. The real audience for this post is OpenAI, not you.
lyu07282 · 1h ago
It adds some pressure, we know now what the bug is about so we can guess which endpoints to poke at, then it's only a matter of time before it leaks. It would be unethical for the researcher to just publish it.
I'm curious which mailbox they sent to, trying to find a mailbox is surprisingly hard even with my Google searching.
requilence · 4h ago
they have security.txt file on their domain and mentioned it in some other place
robswc · 3h ago
Reminds me of a time I found a serious issue with mailgun. Messaged them, no reply. Had to spam their twitter to get a response. Basically you could have stolen tons of API keys from users without their knowledge and mailgun never disclosed it.
I could have actually gone to their office in person if I wanted to be pedantic but it actually seemed like a pretty weird office space lol.
tptacek · 3h ago
I don't think disclosure of reported security issues is really a norm, unless the firm finds evidence the bug was exploited (by someone other than the reporter). It's a good thing to do, but I think the majority of stuff that gets reported everywhere is never disclosed --- with the major and obvious exception of consumer or commercial software that needs to be updated "on prem".
robswc · 3h ago
Makes sense.
The problem I have with it is that there's no way they could have determined if an API key was stolen or not, even to this day.
Basically, their docs (which seemed auto-generated) pointed to a domain they did not own (verified this). So if you ran any API examples you sent your keys to a 3rd party. I know because I did this. There's no way to know that the domain in the docs is simply wrong.
I tried explaining this to the support people, that I needed to talk with a software engineer but they kept stonewalling. I think it was fixed after 24 hours or so.
blibble · 4h ago
good to see more and more hackers refusing to use corporate bug bounty platforms with onerous terms
I certainly wouldn't sign an indefinite NDA for a chance to win:
Average payout: $836.36
openai should be grateful, after all, they want all information to be free
JyB · 4h ago
I believe it is extremely important to disclose that the ‘responses leaks’ you obtained did not originate from LLM models themselves, but rather through other insecure systems / in a more conventional manner.
Just to avoid yet another case of hallucinations outputs getting misinterpreted.
requilence · 4h ago
Right, thank you for the suggestion. Just added a paragraph to the original blog post.
tabletcorry · 3h ago
Your added paragraph appears to suggest the opposite, that this was an LLM response. Was the "leaked data" a response from an LLM directly?
https://jarbon.medium.com/gpt-prompt-bug-94322a96c574
https://snipboard.io/FXOkdK.jpg
I felt like it was a huge deal at the time but it’s surprisingly hard to quickly google it.
Accurate financial data?
How do we know?
What does using not-web-search not having the data have to do with the claim that private chats with the data are being leaked?
???
A lot of AI products straight up have plan text logs available for everyone at the company to view.
I really hope they fix this bug and start taking security more seriously. Trust is everything.
After some hemming and hawing, my most cromulent thought is, having good security posture isn't synonymous with accepting every claim you get from the firehose
Software quality is... Minimal now days.
Mozilla's program, which has been around longer than most, doesn't. Google and Microsoft don't. Meta and Apple don't.
This is water carrying, intentional or not, for a terrible practice that should be shamed, so that it doesn't become standard.
You can shame it all you want, but you can also just publish your bugs directly. Nobody has to use the Bugcrowd platform. You don't even have to wait 45 days; I don't buy these "CERT/CC" rules.
All bets are off with small random startups that do bug bounties because they think they're supposed to (most companies should not run bounties). But that's not OpenAI. Dave Aitel works at OpenAI. They're not trying to stiff you.
Simultaneous discovery (either with other researchers or, even more often, with internal assessments) is super common. What's more, you're not going to get any corroboration or context for them (sets up a crazy bad incentive with bounty seekers, who litigate bounty results endlessly). When you get a weird and unfair-seeming response to a bounty from a big tech company, for the sake of your own sanity (and because you'll probably be right), just assume someone internal found the bug before you did, and you reported it in the (sometimes long) window during which they were fixing it.
A model like GPT-4o can hallucinated responses that are indistinguishable from real user interactions. This is easy to confirm for yourself: just ask it to make one up.
I’m certainly willing to believe OpenAI leaks real user messages, but this is not proof of that claim.
You can spot anyone using AI writing a mile away. It stopped saying "delve" but started saying stuff like "It's not X–it's Y" and "check out the vibes (string of wacky emoji)" constantly.
If the story in OP about getting a company's private financial data is true (i.e. the numbers are correct and nonpublic) that could be a smoking gun.
Either way it's a bad look for OpenAI to have not responded to this. Even if the resolution turns out to be that these are just hallucinations, it should've been investigated and responded to by now if OpenAI actually care about security.
Massive security bug, well spotted. It's like Bank of America showing other people my transactions, or Meta leaking my WhatsApp messages.
This raises some serious questions about security.
For real? At least doesn't match the one on https://keybase.io/requilence
I could have actually gone to their office in person if I wanted to be pedantic but it actually seemed like a pretty weird office space lol.
The problem I have with it is that there's no way they could have determined if an API key was stolen or not, even to this day.
Basically, their docs (which seemed auto-generated) pointed to a domain they did not own (verified this). So if you ran any API examples you sent your keys to a 3rd party. I know because I did this. There's no way to know that the domain in the docs is simply wrong.
I tried explaining this to the support people, that I needed to talk with a software engineer but they kept stonewalling. I think it was fixed after 24 hours or so.
I certainly wouldn't sign an indefinite NDA for a chance to win:
Average payout: $836.36
openai should be grateful, after all, they want all information to be free
Just to avoid yet another case of hallucinations outputs getting misinterpreted.