Приватные чаты пользователей Claude попали в поисковую выдачу Google и Bing
Приватные переписки пользователей с чат-ботом Claude от Anthropic попали в поисковую выдачу Google и Bing — об этом в июле 2026 года сообщил Wired. Диалоги, которыми делились по ссылке, обошли поисковые роботы, и их мог найти любой. Похожая утечка годом ранее затрагивала ChatGPT — тогда OpenAI отключила индексацию общих чатов.
AI-processed from Wired; edited by Hamidun News
Private conversations of Claude users, the AI chatbot from Anthropic, turned out to be accessible in Google and Bing search results — Wired reported this in July 2026. Conversations that people considered private could be found by anyone who typed the right query into a search engine.
What exactly ended up in search
Web pages with Claude dialogues that were crawled by search bots ended up in the Google and Bing index. According to Wired, these were conversations that users perceived as private but that were technically hosted on the open web — which is why they ended up in the results of the two largest search engines.
- The Claude chatbot from Anthropic is affected
- The dialogues were indexed in two search engines at once — Google and Bing
- Wired brought the issue to light in July 2026
- The mechanism is web crawlers scanning publicly accessible pages
- A similar incident a year earlier affected ChatGPT from OpenAI
Why private chats get indexed at all
The reason lies in how the chat-sharing feature is built. When a user generates a link to a conversation with the chatbot, it usually leads to a publicly accessible web page. And anything that sits on the web without an explicit ban on crawling sooner or later comes into the view of search crawlers — the programs that Google and Bing run to populate their index.
Technically, a site owner can block such pages from indexing — via a robots.txt file or a noindex meta tag. But these mechanisms don't always work, and not instantly: a crawler may ignore the directive, and a page that's already been indexed has to be removed from the results separately. Meanwhile the user usually doesn't see that their link ended up in the index at all. Wired stresses that stopping crawlers turns out to be noticeably harder than one might assume.
"This failure shows how hard it is to stop web crawlers from making supposedly private conversations with AI chatbots excessively public,"
Wired's report states.
What this recalls about the ChatGPT story
A similar leak already happened with ChatGPT: in the summer of 2025, dialogues that users shared via link started showing up in Google search. At the time, OpenAI removed the feature that made shared chats indexable and called it a short-lived experiment. The repeat of the same pattern with Claude shows that the AI industry keeps stepping on the same rake with the privacy of shared links — this time at the scale of two search engines, Google and Bing.
What users should do
You can check your own conversations manually: enter a query in Google or Bing using the site: operator plus the address of the chat-sharing service, or paste a fragment of your own dialogue into search. If a chat turns up, you need to revoke its link in the service's settings, and then, if necessary, request removal of the page from the search engine's index through tools like Google Search Console or Bing Webmaster.
The basic rule is simple: don't share conversations containing personal data, work secrets, or anything that shouldn't be publicly accessible via a public link. A link that can be forwarded to another person is, by default, open to a search bot as well — and therefore potentially to everyone else.
What this means
The Claude incident is one more reminder that "sharing a chat" and "keeping a chat private" are not the same thing. As AI services keep expanding their dialogue-sharing features, a significant share of the responsibility for privacy remains with the user: a public link almost always means public access, including for Google and Bing. And chatbot developers themselves are being forced to reconsider which sharing features should even be enabled by default.
Want to stop reading about AI and start using it?
AI News is a curated feed of AI/tech news. Hamidun Academy teaches you to use AI systematically in your work.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.