
Signal views AI agents built into phones as an “existential threat,” and it’s acting accordingly. The encrypted messaging app said Friday that it will block screenshots by default on iOS 27, and on Android when AI assistants try to capture the screen.
The measures keep Apple’s Siri AI and Google’s Gemini Intelligence from snooping on your chats, at least to a degree. The assistants are designed to help users by reading what’s on the screen and pulling in information to better answer their questions. But the byproduct is that chats meant to be private — including encrypted ones — can end up on a remote server outside your or Signal’s control.
The move echoes one from last year, when Signal blocked Microsoft’s Recall from snapshotting chats on Windows 11.
Signal’s concerns aren’t limited to the built-in assistants. It’s also worried about third-party agents like Meta’s Muse, which Reuters caught reading Signal messages via permissions granted during setup. On Android, it read messages from notifications; on Mac, it used accessibility permissions, which Signal warns can be exploited “as a skeleton key.”
The new settings are slightly different on iOS and Android. On iPhone, a new “Block Siri AI” setting is on by default. On Android, Signal uses its existing “Screen security” option. On both platforms, you’ll see a prompt offering to turn the protection off when it detects either a screenshot (iOS) or an AI assistant reading the screen (Android).
But that protection only goes so far. If the other person has switched protections off, blocking screenshots on your end won’t stop an AI assistant or agent from reading your messages on the other person’s device. It also won’t prevent third-party agents from reading notifications or digging through files. Signal says operating system makers don’t offer a way for developers to restrict AI access “without making drastic compromises to the core user experience.” On Mac, that leaves Signal without any user protections at all.
In some ways, this mirrors the problem with AI glasses, which can record bystanders without their knowledge or consent. That’s led to them being banned in some public places, like courtrooms and schools, and being shunned as “perv glasses.” Privacy in the age of AI is becoming less of an individual setting and more of a communal choice, with laws and social shaming being some of the only recourse individuals have.
As for Signal, it wants a standard, system-level way to block AI agents from reading apps’ content. It says it’s talking with Apple and Google and is cautiously optimistic about where that’s headed. Until then, know that “privacy” is a relative term at best.