ChatGPT Users Beware: Your Private Chats Are Showing! Thousands of conversations are turning up in Google search results, often without their creators even realizing it. Some expose personal, sensitive, or even confidential data. Let’s break down the issue and explore how to protect yourself before you’re left red-faced!
We often imagine a total sense of privacy when speaking to an AI like ChatGPT. Many users freely share their thoughts, problems, desires, or memories. However, the reality is now crystal clear.
Recently, OpenAI’s CEO cautioned against using ChatGPT as a therapist. There’s no doctor-patient confidentiality when you chat with this bot. Anything you tell it could potentially be used against you in a legal proceeding!
Only AI tools specifically designed for privacy, like Venice AI or Proton Lumo, can provide a higher level of protection. But even then, there’s no 100% guarantee. This raises important questions.
But concerning ChatGPT, the situation is worse than initially feared. By typing a simple command into Google (site:chatgpt.com/share), users are finding hundreds, even thousands, of conversations!
These aren’t pirate leaks, but rather links made public…by the users themselves. The problem? Many were unaware that these links could be indexed by search engines.
Even worse, they didn’t realize their content would be visible to everyone, without a password, barrier, or anonymization. Over 4,500 publicly accessible discussions have been discovered!
Mental health concerns, relationship disputes, resume rewrites, branding strategies – it’s all there in black and white, ready to surface with a simple click…
If the link is created using the “Create Link” button, it becomes accessible to anyone with the URL. Even worse, if it’s not protected through advanced options, it can be found, scanned, and indexed by search engines like Google or Bing.
What many users didn’t know is that the link could become visible even without intentional action, particularly during an experimental phase launched by OpenAI.
A checkbox was supposed to let users choose whether the link could be discovered via search engines. But the process lacked clarity. This resulted in a massive leak of exchanges that were intended to be private…
Intimate Stories, Sensitive Info, Identifying Details
Most indexed conversations are harmless: cooking recipes, math homework, gardening tips. But a significant portion is far more concerning.
We’re finding accounts of personal trauma, disclosures related to mental health (anxiety, depression, addictions), and questions about relationship or sexual issues…
Some users have also shared detailed professional requests, such as resume or cover letter rewrites. This can severely damage your credibility with potential employers.
The leaks even include business brainstorming sessions about products still in development.
Worse still, some exchanges contain first names, cities, employer names, email addresses… In short, identifying data that allows the author to be traced, willingly or not.
Even if a link is deleted from the ChatGPT interface, it can still remain active in Google’s cache or persist in the search engine’s index for days, even weeks.
OpenAI Backtracks, But It’s Too Late
Alerted by journalists from Fast Company, OpenAI quickly reacted. A few hours after the investigation was published, the company disabled the ability to make “shareable” links visible on Google.
“It was a short-lived experiment”, says a spokesperson. “But it introduced too many risks of accidental sharing”.
From now on, conversations shared via ChatGPT should no longer appear in search engines…at least for new ones. Older ones, however, remain partially indexed until Google cleans up its databases. To accelerate this de-indexing, OpenAI is working with the search engines.
Why Is This a Real Problem?
This embarrassing leak makes us reflect on our new relationship with AI and our skewed perception of confidentiality.
More and more users treat ChatGPT like a digital confidante. A study reported by Techzine indicates that nearly half of American users used an AI chatbot to discuss psychological or personal topics in 2024.
Over 75% mentioned anxiety issues, and 60% discussed mental health or well-being concerns. AI has become our therapists.
And yet, these exchanges are neither protected, encrypted, nor confidential in a legal sense. They can be recovered in the context of an investigation, a hacking, or… a simple Google search.
OpenAI CEO Sam Altman acknowledged: “Users need to understand that what they type can be seen. It’s a tool, not a diary.”
If you used the “Share” function, take a few minutes to clean things up. Here are the recommended steps:
Type into Google: “site:chatgpt.com/share + [your first name, company, personal subject]”. This allows you to see what’s showing up.
Connect to ChatGPT, go to the “Shared Links” tab, and delete anything that’s active.
In the future, avoid sharing conversations containing sensitive information, even indirectly.
Favor screenshots or text copy-pasting if you want to disseminate an AI response without leaving a web trail.
Never treat ChatGPT as a confidential storage space. It’s not a personal drive. It’s not encrypted. It’s not designed for that.
The Internet Doesn’t Forget, Even When You Want It To
This case is a brutal lesson about the illusion of digital privacy. A smooth interface, a reassuring design, an AI that responds kindly – all this gives the impression of an intimate space.
But behind it, there are servers, web protocols, indexing algorithms. And a simple misplaced checkbox can be enough to expose a part of your intimacy to the whole world.
As AI becomes omnipresent in our lives, it’s time to adopt a salutary reflex: consider every line you type as potentially public.
So, since the cat is out, apparently your shared ChatGPT chats aren’t as ”privat” as one would think:
Google dork:
Site:chatgpt.
(.)com/share intext:lootAnd Wayback has 10k+ links, web.archive(.)org/web/*/chatgpt.com/share/*
Time to crape and grep for loot.
Edited since…
— STÖK ✌️ (@stokfredrik) July 31, 2025
Did you accidentally use the sharing feature? Do you tend to distrust AIs, or on the contrary, confide all your secrets to them? Share your opinion in the comments!
<ul class="end-post-layout"><span style="font-size:16px; font-weight: 800; margin-right: 10px; margin-top: 5px;">Partager l'article : </span>
<li class="social-link">
<a href="http://www.facebook.com/sharer.php?u=https://www.lebigdata.fr/chatgpt-vos-discussions-intimes-sont-visibles-sur-google-retirez-les-vite" class="social-link-anchor facebook" target="_blank">
<svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" version="1.1" width="25" height="25" x="0" y="0" viewbox="0 0 512 512" style="enable-background:new 0 0 512 512" xml:space="preserve" class=""><g><path fill-rule="evenodd" d="M255.182 7.758q69.23.79 125.086 34.03a249.734 249.734 0 0 1 88.89 89.434q33.037 56.191 33.825 125.843-1.962 95.3-60.117 162.79c-38.77 44.995-88.425 72.83-139.827 83.501V325.23h48.597l10.99-70h-73.587v-45.848a39.844 39.844 0 0 1 8.474-26.323q8.827-11.253 31.09-11.829h44.436v-61.318q-.957-.308-18.15-2.434a360.743 360.743 0 0 0-39.16-2.434q-44.433.205-70.281 25.068-25.85 24.855-26.409 71.92v53.198h-56v70h56v178.127c-63.115-10.67-112.77-38.506-151.54-83.5S8.691 320.598 7.383 257.065q.785-69.655 33.824-125.843a249.739 249.739 0 0 1 88.891-89.435q55.854-33.233 125.084-34.03z" opacity="1" data-original="#000000" class=""/></g></svg>
<span class="screen-reader-text">Facebook</span>
</a>
</li>
<li class="social-link">
<a href="https://twitter.com/share?url=https://www.lebigdata.fr/chatgpt-vos-discussions-intimes-sont-visibles-sur-google-retirez-les-vite&text=ChatGPT%20:%20vos%20discussions%20sont%20sur%20Google%20!%20Retirez-les%20avant%20qu%E2%80%99elles%20soient%20lues" class="social-link-anchor twitter" target="_blank">
<svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" version="1.1" width="25" height="25" x="0" y="0" viewbox="0 0 1227 1227" style="enable-background:new 0 0 512 512" xml:space="preserve"><g><path d="M613.5 0C274.685 0 0 274.685 0 613.5S274.685 1227 613.5 1227 1227 952.315 1227 613.5 952.315 0 613.5 0z" opacity="1" data-original="#000000"/><path fill="#ffffff" d="m680.617 557.98 262.632-305.288h-62.235L652.97 517.77 470.833 252.692H260.759l275.427 400.844-275.427 320.142h62.239l240.82-279.931 192.35 279.931h210.074L680.601 557.98zM345.423 299.545h95.595l440.024 629.411h-95.595z" opacity="1" data-original="#ffffff"/></g></svg>
<span class="screen-reader-text">Twitter</span>
</a>
</li>
<li class="social-link">
<a href="http://www.linkedin.com/sharing/share-offsite/?url=https://www.lebigdata.fr/chatgpt-vos-discussions-intimes-sont-visibles-sur-google-retirez-les-vite" class="social-link-anchor linkedin" target="_blank">
<svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" version="1.1" width="25" height="25" x="0" y="0" viewbox="0 0 512 512" style="enable-background:new 0 0 512 512" xml:space="preserve" class=""><g><path d="M256 0C114.637 0 0 114.637 0 256s114.637 256 256 256 256-114.637 256-256S397.363 0 256 0zm-74.39 387h-62.348V199.426h62.347zm-31.173-213.188h-.406c-20.922 0-34.453-14.402-34.453-32.402 0-18.406 13.945-32.41 35.274-32.41 21.328 0 34.453 14.004 34.859 32.41 0 18-13.531 32.403-35.274 32.403zM406.423 387h-62.34V286.652c0-25.218-9.027-42.418-31.586-42.418-17.223 0-27.48 11.602-31.988 22.801-1.649 4.008-2.051 9.61-2.051 15.215V387h-62.344s.817-169.977 0-187.574h62.344v26.558c8.285-12.78 23.11-30.96 56.188-30.96 41.02 0 71.777 26.808 71.777 84.421zm0 0" opacity="1" data-original="#000000" class=""/></g></svg>
<span class="screen-reader-text">LinkedIn</span>
</a>
</li>
</ul>
<!-- Display tags -->
<!-- End display tags -->
<p>Notre blog est alimenté par les lecteurs. Lorsque vous achetez via des liens sur notre site, nous pouvons percevoir une commission d’affiliation.</p>
</div><script async src="//platform.twitter.com/widgets.js" charset="utf-8"></script>Worth a look