Signal is tightening its screenshot rules again, this time in response to AI agents built into mobile operating systems. The encrypted messaging app will block screenshots in its iOS 27 app and will stop AI assistants from capturing screen data on Android, Signal President Meredith Whittaker said in a blog post Friday. The change affects Signal users on both Apple and Google platforms.
Whittaker pointed to Google's and Apple's agentic features, including Gemini Intelligence and Siri AI's Onscreen Awareness, as threats to the privacy work third-party developers have built. Those features look at what is displayed on a user's screen to gather context for completing requests. She said sensitive data is captured, stored, and exposed to whichever local database or remote server an operating system-level AI agent is using at that moment, and that Signal loses control of the information once it has been captured. That, she said, endangers everyone taking part in a conversation.
This is not Signal's first move of this kind. The company blocked screenshots last year in response to Microsoft's Recall AI feature, which could capture images of Signal chats on devices running Windows 11. Signal's latest step illustrates how Big Tech's AI strategy reaches downstream to app makers. As AI features have shifted from services outside the operating system, such as chatbots from OpenAI or Anthropic, to deeper integration inside it, third-party apps have had to adjust to conditions that undercut the privacy infrastructure they built.
The timing underscores the stakes. Just last week, researchers at Google DeepMind used AI agents to formally verify Signal's code, building on Signal's own efforts to essentially prove the app's encryption works. That sets a high bar for a security guarantee inside the app, but the guarantee becomes meaningless if an agent can simply take messages off a screen. Samsung is reportedly working on its own agent, Galaxy Claw, and Apple is developing a rival to Claude and ChatGPT as well.
Whittaker acknowledged the restriction is imperfect, noting that AI agents can still reach app data through other routes, such as pulling content already on a user's device via the filesystem. Earlier this week, a Reuters investigation supported by Signal and the AI Now Institute found that Meta's Muse AI agent was using permissions from its initial setup in unexpected ways to reach apps and data a user had not explicitly connected it to. Whittaker said it is a serious problem that Signal and other application developers lack robust tools to stop AI agents from undermining application-level privacy, and that Google, Apple, Microsoft, and others offer no easy way for developers to restrict such access without drastic compromises to the core user experience.
Users who still want to take screenshots can turn the setting off, though Signal's announcement strongly advised against it. Whittaker said doing so would potentially expose Signal content belonging to users and their contacts to these AI agents and would break the company's privacy guarantees.
More AI news from TechManNews.






