Signal is restricting screenshot capabilities on Android and iOS devices to stop artificial intelligence agents from reading user messages and other data. The messaging service stated that agentic AI features built into operating systems are threatening its privacy guarantees.
Protecting User Privacy From AI Agents
Signal President Meredith Whittaker explained in a blog post that operating system features designed to analyze on-screen content undermine the platform’s independent privacy protections. Whittaker emphasized that application developers lack robust tools to prevent AI agents from accessing private data. Tech giants including Google, Apple, and Microsoft fail to provide simple methods for developers to restrict AI data access without harming the core user experience, which is why similar protections cannot yet be implemented on macOS.
To mitigate these concerns, Signal is blocking in-app screenshots on iOS and Android devices, effectively preventing virtual assistants like Siri and Gemini from scanning conversational content. Users retain the option to re-enable screenshots within the app settings, though Whittaker cautioned that doing so exposes private content to AI tools and compromises security standards. Whittaker also noted that blocking screenshots serves as an incomplete fix because it does not stop AI agents from retrieving data through alternative channels such as the device file system.
Calls for Greater Developer Control
Signal previously implemented similar measures to shield secured chats from Microsoft’s controversial Recall feature. According to the encrypted messaging provider, developing comprehensive solutions remains difficult due to insufficient documentation regarding how AI systems access application data. Consequently, the organization is urging Apple, Google, and Microsoft to supply straightforward technical tools and accurate documentation for developers.
Furthermore, Signal highlighted the need for safeguards against third-party tools, such as Meta’s Muse, which can exploit deep accessibility permissions within applications to extract sensitive information. Whittaker warned that integrating system-level AI agents without proper safeguards poses a severe threat to application privacy and affects independent developers across the entire digital ecosystem.








