AI · 21 November 2026
Signal Blocks Screenshots on iOS 27 and Android to Stop AI Agents
Signal is blocking screenshots of conversations on iOS 27 and Android to stop AI agents such as Siri, Gemini Intelligence and Meta's Muse from reading private chats. The company concedes its protections are incomplete, with gaps remaining on macOS.
What happened
Signal, the encrypted messaging app, is blocking screenshots of conversations on iOS 27 and Android so that AI agents cannot read private chats. The named targets include Apple's Siri AI, Google's Gemini Intelligence and Meta's Muse, which can otherwise read what appears on screen.
Signal describes the risk as an "existential threat" to privacy. It also concedes that its protections are incomplete, and that gaps remain on macOS. The reporting by The Decoder argues that Apple, Google and Microsoft should take similar steps, though the broader industry response has not yet been set out.
Why it matters
The move highlights a shift in what "private" means on a personal device. Once an AI agent can read whatever is on screen, the privacy boundary is no longer just the app's encryption. It also depends on the operating system, the assistant and the permissions a user has granted, often without fully understanding the trade-off. For organisations building digital services, this is a design question as much as a security one: when a user asks an assistant to summarise a conversation, what have they consented to, and who else sees the result?
The story also shows how quickly AI features are changing user expectations. Users may enjoy the convenience of an assistant that understands their context, but they are increasingly likely to ask where that context goes. Products that make these choices visible and reversible are likely to earn more durable trust than those that treat data access as a default.
The Renascence take
Most coverage frames this as a fight between a privacy-focused app and large platform companies. The more useful reading is that users are being asked to make consent decisions they cannot see. Signal's admission that its protections are incomplete is itself a useful signal: no single app can set the boundary for an entire device.
The real lesson is that privacy has become an experience problem, not just a technical one. Users who cannot tell what an AI agent can see will either overshare or switch off the feature entirely, and both outcomes damage the relationship. Operators should map where their own products expose conversation or screen content to AI features, and make those flows explicit, reversible and easy to understand. Trust is built at the moment of permission, not in the privacy policy.
Sources
This briefing was written by our Newsdesk, synthesising reporting from the outlets below. Follow the links for the original coverage.
FAQ
Questions we get on this topic
More in AI
Stay ahead of CX
Get the signal, not the noise.
The stories shaping customer experience — plus the Journal and Experience Loom — in your inbox.


