Shared ChatGPT conversations and Claude artifacts have both been found in search results. Paste a link to build the search query that checks yours.
Runs entirely in your browser — the link you paste is never sent anywhere.
Nothing is sent anywhere — the queries are built in your browser and open in a normal search tab.
Twice now, an "anyone with the link" share page from a major AI tool has turned out to mean "anyone with a search engine". OpenAI had it first with shared ChatGPT conversations; in late July 2026 the same thing surfaced for shared Claude chats and artifacts, with reports of indexed pages containing health information, internal company documents, and personal contact details. Both were fixed after the fact, but a page that was crawled while the gap was open can persist in search results afterwards. This tool does not guess: it builds the exact site: query for the link you paste so you can check the live index yourself, and tells you where to revoke the share if something comes back.
Paste the share URL you want to check — a chatgpt.com/share, chat.openai.com/share, or claude.ai/share address.
Open the narrow query first: it searches for that exact page. Then try the broad one to see the wider share path for that service.
If your page appears, revoke the share in the original tool. That removes access; the search listing drops out on the next crawl.
Suppose you shared a Claude conversation in June to show a colleague how you had worked through a contract, and you have not thought about it since. Paste that share URL and the checker identifies which service it belongs to, then produces two searches: a narrow one for that exact page, and a broader one across the whole share path for the service. Run the narrow one first. If the page appears, it has been crawled, and deleting the share is what actually removes public access — the search result itself will drop out once the crawler revisits and finds nothing. The broad query is worth running once even if your own page is clean, because seeing how many strangers’ conversations are sitting there is the fastest way to understand why share-link privacy is not something to take on trust. Nothing you paste here leaves your browser — the queries are assembled locally and open in a normal search tab.
No. The parsing and query building happen entirely in your browser; the link is never transmitted to a server here. When you run a check, you are opening a normal search-engine tab with a query in the URL — the same thing you could type by hand, just assembled correctly for the service the link belongs to.
Revoke the share in the tool that created it — in ChatGPT that is the shared-links list in settings, and in Claude it is the share control on the conversation or artifact itself. Deleting the share is what removes public access; the search listing then falls out on its own when the crawler next visits and finds the page gone. If it needs to disappear faster, both Google and Bing offer a removal request form for content that is already unavailable.
It means that page is not in that search index today, which is genuinely good news but is not a guarantee. Indexes lag, coverage differs between search engines, and a page can be crawled later. The durable fix is not to audit repeatedly but to stop putting anything sensitive behind a link whose indexing behaviour is somebody else’s decision.
Published documents are served with noindex directives in two independent places — an X-Robots-Tag response header and the page metadata — so a crawler reading either one is told to skip the page. They are also served from a separate cookieless origin from the app, and expire after seven days on the free tier — both without paying anything. Password protection, for when the URL alone should not be enough, is a paid feature. The part the incidents above were missing is the free part: a share page that tells crawlers to stay away from the moment it exists.
Links gratuitos duram 7 dias com uma conta gratuita. Em um plano pago, cada página que você publica fica online permanentemente.
Ver preços