OpenAI has withdrawn an experimental setting that allowed shared ChatGPT conversations to be indexed by Google and other search engines after media reports revealed thousands of sensitive chats—ranging from trade secrets to quasi-therapy sessions—were appearing in public results. The opt-in feature added a small checkbox labelled “Make this chat discoverable” when users generated shareable links. Although participation required two explicit clicks, critics said the wording was unclear and encouraged accidental oversharing. Search-engine observers counted roughly 5,000 exposed threads before OpenAI intervened. Chief Information Security Officer Dane Stuckey called the tool a “short-lived experiment” and said OpenAI is working with Google, Bing and DuckDuckGo to purge indexed content, a process scheduled to finish by 2 August. Stuckey acknowledged that the setting “introduced too many opportunities for folks to accidentally share things they didn’t intend to.” The episode follows CEO Sam Altman’s public reminder that ChatGPT exchanges lack doctor-, lawyer- or therapist-style confidentiality and can be subpoenaed. It underscores the privacy risks generative-AI platforms face as usage widens and regulators scrutinise data-handling practices.