Signal is limiting screenshots to keep AI agents from reading private chats. On iOS 27, the app blocks screenshots; on supported versions of Android, it enables protection when an AI assistant tries to access the screen. Signal told Reuters that AI access to personal data poses an “existential threat” to privacy.
Signal limits screenshots to protect chats from AI agents
How AI accesses messages
System-level AI tools, including Apple’s Siri AI and Google’s Gemini Intelligence, can read screen contents to carry out user tasks. End-to-end encryption does not protect messages from being processed after they are decrypted: an AI agent can read them directly from the display and send them for external processing.
Recently, Meta’s Muse AI agent accessed Signal content on Android and macOS thanks to broad permissions granted during setup. The capabilities it used included accessibility tools and screenshot capture on macOS. Meta said this happened only with the user’s explicit permission.
Protection is not yet possible everywhere
Signal has already blocked Microsoft Recall on Windows, but considers existing measures incomplete. On macOS, screenshots cannot be reliably prevented, while AI access to the file system and notifications continues to pose a risk.
The company is urging Apple, Google and Microsoft to make it easier to completely block AI agents from accessing apps. Dataist’s take: Signal’s restrictions reduce the risk of chats being read from the screen, but do not close other routes to data through files and notifications.
Daily AI news
Every day we pick what actually matters in AI and explain it plainly — no hype, no filler. Subscribe if you want to follow where the industry is going.
Only what matters — every day
Follow on X