Meta Muse Leaked Secrets When Users Shared Chats

Meta has suspended a sharing feature in its Muse AI after reports that users' personal and business secrets were exposed when they shared conversations, according to Ars Technica. The company is investigating the privacy issue. The incident highlights a risk that grows as AI assistants become embedded in everyday work: chat logs can contain sensitive details such as credentials, client information, internal strategy, and proprietary business context. A feature designed to make conversations social or collaborative can turn private prompts into unintended disclosures. The problem is not limited to one product. Many AI tools encourage users to paste documents, ask for code fixes, summarize meetings, or brainstorm business plans. Those interactions often include data that was never meant to leave an organization. When sharing features are added, users may not understand exactly what will be visible, who can access it, or how long it will remain available. Meta's decision to pause the feature is a short-term containment step, but it also points to a larger governance challenge. AI companies need clear defaults, strong warnings, and granular controls for sharing. Organizations using AI assistants need policies for what can be entered into chat systems and how logs are stored. The Muse episode is a reminder that convenience features can create security incidents. Before sharing becomes a standard part of AI chat, vendors will need to show that they can protect the sensitive material users already bring into these systems.