
Photo: Kritzolina, 2023, via Wikimedia Commons (CC BY-SA 4.0)
If you tried to screenshot a Signal chat this weekend and got nothing, that is on purpose. On Friday, Nov. 20, 2026, Signal began blocking screenshots by default in its iPhone app on iOS 27, and switching on the same protection in its Android app whenever it catches an AI assistant trying to read the screen. Signal president Meredith Whittaker called the AI agents now built into phone operating systems "an existential threat" to the app’s privacy promise.
The short answer: Apple’s Siri AI and Google’s Gemini can now look at whatever is on your screen to answer questions and take actions. Signal’s encryption protects messages on their way between phones, but not once they are displayed. The only lever Signal has to keep those assistants out is the same one apps use to block screenshots, so it pulled it. You can switch the block off in a few taps, but Signal warns that doing so can expose your chats, and the chats of the people you talk to, to those assistants.
Here is what changed, how to turn it on or off on each device, why a Mac can’t get the same protection yet, and what the fix still doesn’t cover.
What changed on Friday
Signal announced the change in a post on X at 1:31 a.m. Pacific time on Nov. 20, saying it was "rolling out a new default to restrict screenshots in the Signal app" and that users could re-enable screenshots in settings. By Saturday night that post had been seen more than 113,000 times. Whittaker’s longer blog post went up the same day. The Verge covered it within hours, and TechCrunch, Fast Company and others followed.
According to Signal’s support page, the protection works in Signal 8.37 or later on iPhones running iOS 27 or later, and in Signal 8.35.1 or later on Android. While it is on, it blocks three things:
- Screenshots and screen recordings taken inside Signal.
- Previews of your chats in the iPhone App Switcher or Android’s recent-apps list.
- On-device AI assistants reading what Signal is showing on screen.
It does not change how messages are encrypted, who can read them or any other Signal setting. The change is only about what can be captured from your own display.
How to turn Signal’s screenshot block off (or on)
The setting has a different name, and behaves a little differently, on each platform.
On iPhone (iOS 27)
On an iPhone the setting is called Block Siri AI, and it is on by default after you update.
- Open Signal, tap your profile picture and choose Settings.
- Tap Privacy.
- Tap Block Siri AI and pick On, Off, or Off for 10 Minutes.
The 10-minute option is the one most people will want. It lets you grab a screenshot, for example to save a receipt or report an abusive message, and then turns the protection back on by itself. If you try to take a screenshot while the block is on, Signal shows a message that it was prevented, with an Options button that offers the same three choices. If you turn the block off, whether your chats show up in the App Switcher is decided by the separate Hide Screen in App Switcher setting.
On Android
On Android the setting is the older Screen security switch, and Signal does not flip it for everyone at once. It turns on automatically the first time Signal detects an AI assistant trying to read its screen, and Signal shows a message explaining what happened.
- Open Signal and go to Settings.
- Tap Privacy, then App security.
- Toggle Screen security on or off.
There is one catch. If you turn it off from the pop-up that appears when Signal blocks an assistant, Signal will turn it back on after 90 days. If you turn it off manually in settings, Signal won’t remind you. Turning Screen security on also hides Signal in the recent-apps view. While it is on, Android simply refuses to take the screenshot.
Why the different approach? In a footnote, Signal explains that Android comes in many versions from many phone makers, and on many of them the app has no reliable way to tell that a screenshot was attempted once screenshots are disabled. Rather than have screenshots silently stop working and confuse people, it chose to watch for an AI assistant reaching for the screen and react to that.
On a Mac
Signal can’t block screenshots on macOS, and says so plainly. Apple doesn’t give developers a documented way to stop screenshots of an ordinary app window. The screen-protection tools that do exist on the Mac are built for streaming video and other DRM-protected content, and Signal says it found "a minefield of bewildering decisions that place user privacy at risk" when it tried.
Its advice for macOS 27 users is to add Signal to the Excluded Apps list in System Settings > Siri. Apple added that list in its 27.2 software updates, and the Electronic Frontier Foundation, which had pushed for it, walked through it in October. On an iPhone the same list sits under Settings > Siri > Excluded Apps. Once an app is excluded, asking Siri about what is on screen in that app simply doesn’t work.
On Windows
Windows users already have this. In May 2025 Signal Desktop turned on Screen security by default on Windows 11 to keep Microsoft’s Recall feature, which takes snapshots of the screen every few seconds and makes them searchable, from capturing Signal chats. It did that by marking its window the way video apps mark protected movies, so that the content doesn’t show up in Recall or any other screenshot tool. The new mobile change applies the same idea to phones.
As an Amazon Associate, Aglena earns from qualifying purchases.
Why Signal says AI assistants are the problem

Photo: Jakub CA, 2025, via Wikimedia Commons (CC BY 4.0)
The trigger is a new generation of assistants that can see the screen. Apple released Siri AI in beta on Sept. 14, 2026, with what it calls onscreen awareness: Siri can answer questions or take actions based on whatever app you are looking at. On Android, Gemini’s screen context feature can use text from the screen or a screenshot as context for a request, and Gemini can also read notifications if its Device Assistance connection is on.
Signal’s encryption was built for a world where the threat sat between two phones. Messages are scrambled from the moment they leave one device until they arrive on the other, and Signal’s code and protocol have been picked over by security researchers for more than a decade. An assistant that reads the screen sidesteps all of that, because it sees the message after it has been decrypted and displayed. What happens to that text next, whether it stays on the phone or goes to a server, is up to the operating system, not Signal.
We all need to recognize that the current integration of AI agents into the operating system, without these common-sense remedies, represents an existential threat to privacy at the application layer, and the ability of independent developers like Signal to protect your sensitive data.
Meredith Whittaker, president of Signal, on the Signal blog, Nov. 20, 2026
This isn’t only theoretical. Two days before Signal’s announcement, Reuters published tests showing that Meta’s Muse AI agent could pull Signal message content through Android notifications and, on a Mac, through accessibility permissions and screenshots, as Notebookcheck summarized. Apple told Reuters an operating-system update had closed the particular Mac route that was tested. Signal’s post singles out third-party agents like Muse, which ask for deep accessibility permissions during setup. Those settings were meant to help people use their apps, Signal wrote, "not to act as a skeleton key that provides access to AI agents."
It is the same tension that runs through every agent product right now. The features that make an assistant useful, like reading your messages, filling in forms or switching between apps for you, are the same permissions that make it dangerous in the wrong hands, which is why security teams have started worrying about AI hacking agents as much as about the assistants themselves.
What the fix still doesn’t cover
Signal calls its own remedy "incomplete," and it is worth being clear about the gaps.
- Files already on the phone. Blocking screenshots doesn’t stop an agent from reading app data through the filesystem or other back doors. Signal names this directly.
- Notifications. If your notifications show message previews, an assistant that can read notifications may see them before you open the app. Setting Signal’s notifications to show only the sender, or nothing, closes that route.
- Macs and some Android phones. macOS has no automatic protection at all, and Android detection doesn’t work reliably on every manufacturer’s version of the system.
- The other person’s phone. The setting protects your own screen. If someone you talk to turns it off on their phone, your messages on their display are as exposed as theirs.
- Legitimate screenshots. People screenshot chats to save addresses, keep evidence of harassment or share a funny moment. Signal’s 2025 Windows post noted that some accessibility tools, like screen readers and magnifiers, may also not work properly with screen security on, which is why it kept the switch easy to find.
An assistant that reads the screen sidesteps all of that, because it sees the message after it has been decrypted and displayed.
What Signal wants from Apple, Google and Microsoft
Signal’s complaint is less about any single feature than about missing controls. It says it has spent months doing "forensic investigations" on Android and iOS just to learn what these assistants actually access, because the documentation isn’t there. It is asking the three big operating-system makers for two things.
- An opt-out for apps. A simple, documented way for a developer to keep its app out of AI data access, covering screen reading, files, notifications and other system hooks, without breaking the app.
- Real documentation. Clear technical specifications for how these agents collect and process data, so apps like Signal can tell users what is and isn’t protected.
The tone wasn’t all combative. Signal said it is "cautiously optimistic" that Google and Apple are engaging with it on how these systems should handle privacy-sensitive apps, and that it hopes to have more to share soon. Apple’s per-app Siri exclusion list, added after pressure from groups like the EFF, shows the kind of switch Signal wants, only built for developers to set rather than leaving it to each user to find.
Should you leave it on?
For most people, yes. Leaving the block on costs you the ability to screenshot a chat without an extra tap, and the 10-minute option on iPhone covers most real needs. Turning it off doesn’t stop Signal from encrypting your messages, but it does let the assistant on your phone see whatever you have open, and Signal can’t promise where that text goes next.
If you rely on screenshots or accessibility tools, turn it off knowingly, and consider the other steps above: hide message text in notifications, and on an iPhone or Mac add Signal to Siri’s Excluded Apps list. Since the same assistants will keep gaining new ways to plug into apps, expect this to be the first of several settings like it, from Signal and from other apps that handle private data.
Further reading

The Age of Surveillance Capitalism
Shoshana Zuboff’s landmark account of how tech companies turned personal data into a business, and why the fight over who sees your screen keeps coming back.

Privacy Is Power
Oxford philosopher Carissa Véliz on why privacy is collective, not just personal, with practical steps for taking back control of your data.

3M Privacy Filter for 14-Inch Laptops
A low-tech fix for over-the-shoulder snooping: the screen blacks out from side angles and flips up when you want to share.

Logitech Brio 101 Full HD Webcam
A 1080p webcam with a built-in sliding privacy shutter, so the camera is physically covered when you’re not on a call.

Apple AirTag 4-Pack
Apple’s item trackers use the Find My network, which encrypts location reports end to end, and iPhones warn you if an unknown AirTag is traveling with you.
