Categories

AI

Signal Adds Screenshot Protections Against Smartphone AI Assistants

Signal says it will restrict screenshots on iOS 27 and enable screen security on Android when it detects an AI assistant accessing a chat.

The Signal messaging app on an Android phone
Image credit: ShreyasMinocha / Wikimedia Commons, CC0
The Signal app on Android.

Signal is adding protections intended to stop smartphone AI assistants from scanning and storing private conversations in its messaging app. The company says it will restrict screenshots for the latest agentic AI features built into iOS 27 and automatically enable a screen-security setting on Android when it detects an assistant trying to read a Signal screen.

Signal president Meredith Whittaker called operating-system-level AI integrations an “existential threat” to the app’s mission. She said Apple, Google and Microsoft are changing the environment apps operate in ways that make it increasingly difficult to protect privacy.

How the protections work

Signal’s concern is that AI assistants can scan what appears on a device’s screen. If an assistant captures a Signal conversation, the data may be sent to a local database or remote server used by that system, outside the messaging app’s control and encryption protections.

Users who want to take screenshots can turn off the protection in Signal’s settings. The company warns that doing so could expose both their messages and those of their contacts to AI agents.

The move follows Signal’s earlier effort to prevent Microsoft’s Recall feature from periodically capturing images of its messages on Windows. The company said the new measures address AI features built into phones, as well as third-party agents that may access device content.

Limits and concerns about third-party agents

Signal said the protections cannot prevent every form of access. They do not stop an agent from retrieving data directly from a device’s filesystem, and Signal has not been able to implement the measures on macOS because, it says, Apple does not provide documented tools to block screenshots reliably.

The company also raised concerns about agents such as Meta’s Muse, which can use accessibility permissions intended to help people. A journalist found that Muse had accessed private iMessages despite his settings, then gave an inaccurate explanation of how it knew their contents.

Signal is asking Apple, Google and Microsoft to provide developers with a straightforward, documented way to prevent apps’ data from being accessed by AI systems. It also wants clearer specifications on how those systems handle information, saying it has spent months investigating their behaviour.

Sources and references

  1. Decrypt – Signal Calls OS-Level AI an 'Existential Threat' and Blocks Screenshots to Fight Itdecrypt.co

Share this story

All AI

Choose which cookies Clore may use. You can change this at any time from the footer.

Necessary

Remember your theme and cookie choices, and keep editors signed in. These cannot be switched off.

Always active

Analytics

Help us understand which stories are read so we can improve coverage.

Marketing

Allow advertising partners to show more relevant ads.