Signal has introduced new restrictions on mobile screenshot access to prevent AI agents from reading private user messages. These settings are enabled by default, and users will receive a pop-up notification if the application detects an attempt to capture the screen or if an AI assistant tries to access the interface.

The company stated that the integration of AI directly into operating systems, such as Apple's Siri and Google's Gemini, poses a significant threat to user privacy. Signal argues that these agents can bypass existing security measures by extracting information from the screen and transmitting it to external servers, effectively undermining end-to-end encryption.

Signal is calling on major technology companies, including Apple, Alphabet, and Microsoft, to provide developers with more effective tools to block AI agents from accessing sensitive data. The company demands greater transparency regarding how these AI agents operate, allowing developers to assess data access and adjust their systems accordingly.

The move follows reports that Meta's Muse AI agent could bypass security by utilizing accessibility permissions on macOS and Android to capture and extract text from Signal chats. While Apple recently updated its system to block this specific method, it simultaneously introduced new permissions for AI agents that provide similar capabilities with additional user warnings.

Signal emphasizes that developers need a direct mechanism to opt out of AI data processing, particularly concerning screen reading and file system access. As a non-profit organization, Signal maintains that its mission to provide secure communication is currently under threat due to the deep integration of AI agents within consumer device operating systems.