📰 Key Highlights

Claude’s shared chats and Artifacts feature had a privacy leak over the weekend. Reddit users discovered that searching “site:claude.ai/share” on Google surfaced numerous share links that should have been private, with some content including medical records, internal company documents, and children’s names and phone numbers. The issue appears to stem from Claude’s “shared chat” feature, which generates links labeled “anyone with the link can view” — the intent was to let users share with friends, colleagues, or small groups, not to be publicly indexed by the whole web. By comparison, Google Docs’ similar sharing feature doesn’t get publicly indexed by search engines. When responding, Anthropic leaned toward blaming users, saying share links only show up in search results when they’re pasted somewhere publicly visible like forums or social media, and that privately sent links won’t be indexed. Spokesperson Amie Rotherham also emphasized that the links themselves can’t be guessed or discovered, and that the company doesn’t provide conversation directories or sitemaps to Google and other search engines. Before the fix was rolled out, Futurism reported that the leaked content included detailed medical records of real patients, clinical trial results containing patient names, elementary school children’s names and phone numbers, documents marked “for internal use only,” and performance reviews containing employees’ personal info. The leaked Artifacts also included code and work notes. Fortune further pointed out that in at least one conversation labeled “shared by Anthropic,” Claude generated pornographic content — which directly violates Anthropic’s usage policy that explicitly prohibits generating explicit sexual content. It’s still unclear how that content was induced. As of Monday afternoon, TechCrunch’s testing of the same search method could no longer find relevant results, showing the issue has been partially fixed.

💬 JudyAI Lab Perspective

Anthropic’s Claude had a privacy leak over the weekend with its “shared chat” feature — a quick Google search surfaced medical records, internal documents, and children’s personal info that should have stayed private. The issue has drawn attention.

What AI builders should take away from this is the design of default visibility for share links. Claude’s share links are labeled “anyone with the link can view,” but in practice they got indexed by search engines — inconsistent with how Google Docs and similar features behave. This shows how easy it is for users to misunderstand the line between “link sharing” and “public indexing” in product design. When a feature’s default behavior doesn’t match user intuition, even if “the link can’t be guessed,” the risk gets amplified the moment it’s posted on a public page where crawlers can grab it. This is a reminder: when building any product with sharing functionality, the default state of “who can see this” must be clearly visible — not left for the user to figure out.

If your product also has a sharing feature, now’s the time to check whether your share links can be indexed by search engines, and confirm the default permissions are conservative enough.

📅 Source Info

🔗 Further Reading