Signal has updated its security measures to block screenshots on iOS 27 and restrict access on supported Android versions when an artificial intelligence assistant attempts to view the screen. The messaging app warned regulators that OS-level AI tools pose an existential threat to privacy by allowing agents to capture decrypted messages directly from the display. Systems like Apple’s Siri AI and Google’s Gemini Intelligence can utilise screen content to perform tasks, which undermines end-to-end encryption by processing data externally without the user’s active consent.
This restriction follows a recent incident where Meta’s AI agent, Muse, accessed Signal content on Android and macOS through broad permissions granted for accessibility tools. While Meta states this occurred only with explicit user permission, Signal argues current protections remain incomplete because macOS screenshots cannot be reliably blocked and AI access to file systems remains a risk. The company is now urging Apple, Google, and Microsoft to simplify the process for users to block AI agents from applications entirely.
- Protection applies to iOS 27 and specific Android versions
- Microsoft Recall on Windows was previously blocked
- macOS and file system access remain unaddressed risks




