Signal Blocks Screenshots to Limit AI Access to Private Chats
publishers 23articles 23first reported 20 Nov, 09:30 UTCdeveloping since 20 Nov · 3 editions
Signal is introducing protections on Android and iOS 27 intended to prevent operating-system AI assistants from reading private conversations displayed in its app.
On iOS 27, a new “Block Siri AI” setting will be enabled by default. On Android, Signal will activate its existing Screen Security feature when it detects an AI assistant trying to read the screen. The measures also prevent users from taking screenshots inside Signal, though they can be disabled in the app’s settings.
Signal President Meredith Whittaker said AI agents integrated into operating systems are undermining the messaging service’s ability to guarantee privacy. Turning off the protections could expose both a user’s messages and their contacts’ messages to AI tools, she warned.
The company acknowledged that blocking screenshots is an incomplete safeguard. It does not stop agents from obtaining information through notifications, local files or broad accessibility permissions, and protections disabled on one participant’s device cannot protect messages viewed there. Signal also said it cannot offer equivalent safeguards on macOS because operating-system providers do not give developers sufficient control over AI access.
Signal previously used screenshot restrictions to prevent Microsoft’s Recall feature from capturing conversations on Windows 11. Its latest changes follow broader concerns about agents such as Meta’s Muse, which has accessed Signal content through notification and accessibility permissions during testing.
Signal called on Apple, Google and Microsoft to provide developers with clearer documentation and system-level tools for restricting AI agents without impairing core app functions.
HOW THIS STORY WAS MADE
Written from 8 articles, 4 with the publisher's own text; 8 independent newsrooms once syndicated copies count as one; this version written 30 h after the record first saw the story; the editor kept it as written.
- Publishers
- 23
- Source articles
- 23
- Given to the writer
- 8, 4 with the article's own text
- Independent newsrooms
- 8
- This version written
- 30 h after the record first saw the story
- Second model (editor)
- kept as written
- Publication gate
- passed
- Human review
- none
WHAT THE EDITOR DID
The second model checked the writer's draft against the same sources and published it as written.
Written by a language model from the sources above, then checked by a second model that may only cut, attribute or correct. How it works →