The short version
A share link turns a private conversation into a page that anyone with the address can read. In July 2025, ChatGPT conversations whose owners had ticked "Make this chat discoverable" were found through Google, and OpenAI withdrew the option the next day. A link to a conversation holding a client's facts is a copy of those facts, kept by the provider, readable by whoever has the address.
What it is
Chatbots such as ChatGPT (whose "Share public link to chat" window The Register described on 1 August 2025) and Claude (checked 1 October 2026) let a user share a conversation by creating a link to it. The link is used to show a colleague an answer, to pass a working prompt to a trainee, or to keep a record outside the tool. It carries whatever the conversation holds: the facts typed in, the text of a pasted document, the draft that came back.
Two things are easy to miss. The first is who can open it. Anthropic's help page on public links for shared chats, checked on 1 October 2026, says: "Anyone with the link can open it. No Claude account is required." The second is where the link goes after it is sent. A link pasted into an email or a group chat can be forwarded, and a link posted anywhere public can be found.
A sighting
On 30 July 2025, Fast Company published a report by Chris Stokel-Walker that Google was indexing ChatGPT conversations that users had shared. ChatGPT's share dialogue had offered a box to tick, "Make this chat discoverable", which allowed a shared conversation to appear in web searches. The report, as syndicated by Fast Company Middle East on 31 July 2025, said that "Nearly 4,500 conversations come up in results for the Google site search", and that Google and OpenAI had declined to comment.
On 31 July 2025, OpenAI's Chief Information Security Officer, Dane Stuckey, announced on X that OpenAI had removed the feature. He called it "a short-lived experiment to help people discover useful conversations", said that it had required users to opt in, and gave the reason for removing it: it "introduced too many opportunities for folks to accidentally share things they didn't intend to". He said OpenAI was also working to remove indexed content from the search engines concerned.
The Register reported on 1 August 2025 that the option had gone, and that searches on other engines were still returning results from ChatGPT's shared pages after the announcement.
The example below is made up to show the pattern. An associate asks a chatbot to summarise a client's witness statement, then posts the conversation's share link in the matter's group chat so the supervising partner can read the summary. The partner forwards the link to counsel's clerk. The conversation, with the statement's contents in it, is now readable by anyone who has that message.
Why it happens
A share link looks like a link to a document on the firm's own system, where access follows the matter. It is not: it is a page on the provider's site, and access follows the address. The discoverable box made the step from shared to searchable a single tick, and OpenAI's stated reason for removing it was the risk of people sharing things by accident.
What to say back
"Who else can open that link, and what is in the conversation behind it? If it holds client facts, can the link be switched off and the summary sent through the firm's own system instead?"
Origin
- 30 July 2025. Fast Company reports that shared ChatGPT conversations marked discoverable are appearing in Google results.
- 31 July 2025. OpenAI's Chief Information Security Officer announces the removal of the discoverable option.
- 1 August 2025. The Register reports the option removed and shared pages still turning up on other search engines.
- Now. Share links themselves remain. On 1 October 2026, Anthropic's page said that its public links carry a "noindex" instruction asking search engines to leave them out, but that anyone with a link can open it, without an account. A share link can be switched off; whether a provider keeps a copy after that is a question for its terms.