

Signal, the encrypted messaging app, is adding protections to limit AI assistants' access to private conversations displayed on users' screens. The feature is rolling out on Android and iOS 27. It targets Google’s Gemini and Apple’s Siri, which can analyze content shown on a phone screen.
The protections are designed to prevent AI tools from capturing sensitive messages displayed on users' screens. It will enable Screen Security on Android when it detects an AI assistant trying to read the screen. On iOS 27, it is introducing a Block Siri AI setting.
Signal says the new protections also prevent users from taking screenshots inside the app. Users can disable these safeguards through Signal’s settings if they prefer.
The company says end-to-end encryption protects messages during transmission, but it cannot prevent an AI assistant from capturing information displayed on a device's screen.
Gemini can analyze on-screen content without requiring users to share screenshots manually. Siri also has features that let it understand content displayed on an iPhone.
Signal says these capabilities create a privacy risk when an AI assistant reads a private conversation that was never meant to leave the app.
“Signal is calling on Google, Apple, and Microsoft to give app developers better tools to restrict AI access without compromising app functionality,” the company said in its announcement.
Also Read: How to Evaluate AI Assistants for Data Science Tasks
This is not the first time Signal has pushed back against AI-driven screen capture. Last year, it blocked Microsoft’s Recall feature from taking screenshots of conversations on Windows 11.
The company warned that even disappearing messages could remain in Recall’s searchable database beyond their intended lifespan.
Signal admits its current fix has limits. It cannot reliably offer the same protection on macOS, and it may not work on every Android device. Limiting screenshots also does not stop AI agents from reaching sensitive information through other paths, such as local files.
Signal is asking platform makers to provide clearer documentation and stronger controls for restricting AI access. It believes developers should be able to decide where AI access ends without losing core app features.
The issue extends beyond messaging apps, as banking, healthcare, and workplace apps may also hold information users would not want an AI assistant to read.