Operator of Claude.ai; built the share-chat and Artifacts features whose pages lacked noindex tags, causing the exposure, then added noindex tags and issued a statement defending its sharing controls after the incident broke.
Search engine that crawled and indexed the unprotected share pages; began removing them from results within roughly a day and issued a statement disclaiming responsibility for what pages a site owner makes public.
RE
Reddit users and a GitHub archivist
A Reddit post first surfaced the site:claude.ai/share search technique and gained roughly 4,000 upvotes; a GitHub repository named Shared-Claude-Chats then archived at least 11,241 exposed messages before links were pulled.
Faced a nearly identical 2025 incident via an opt-in discoverability checkbox that let Google index over 4,500 shared ChatGPT conversations; OpenAI removed the toggle entirely afterward.
Grok's share feature generated public, fully crawlable URLs by default with no disclaimer, resulting in a far larger-scale exposure than Claude or ChatGPT.
Individuals whose medical records, clinical trial data, children's personal information, crypto wallet seed phrases, and internal company documents were exposed to public search, often without realizing sharing implied search-engine discoverability.