The Hacker News
Signal blocks AI agents from reading private chats on phones
Signal has announced it will block screenshots by default to stop AI agents from reading private chats on Android and iOS, though it cautions that the protection is incomplete. The feature applies to agent‑enabled systems in iOS 27 and on Android versions where the encrypted messaging app can detect an AI assistant attempting to read the Signal screen.
It builds on a Windows 11 setting Signal introduced last year after Microsoft’s Recall feature, and comes as Meredith Whittaker, president of the Signal Foundation, warns that OS AI agents scanning screen content undermine Signal’s privacy guarantees.
Users can disable the option by going to Settings > Privacy > App Security and toggling off “Block Siri AI” on iOS or “Screen security” on Android. Signal notes that turning this off may expose Signal content to AI agents and weaken privacy guarantees. The foundation says its remedy is incomplete, acknowledging that other methods—such as accessing content already on a user’s device via the filesystem—could still enable AI agents to retrieve data.
It also criticises developers for lacking robust tools and documentation to prevent third‑party agents from accessing sensitive app data, and urges big tech firms to provide an easy opt‑out for apps and clearer technical specifications on how AI agents function, particularly regarding data access and processing.
The situation follows Reuters reporting that Meta’s Muse AI agent bypassed official tools to access Signal messages on Mac and Android by abusing broad OS permissions during setup, underscoring broader concerns about AI agents and privacy. Signal cites this to advocate for clearer controls and documentation, including easier opt-out mechanisms and improved guidance from platform providers.