Why Signal is blocking AI assistants from reading your screen

Signal is restricting screenshots in its Android and iOS apps so that AI agents built into phone operating systems, such as Gemini and Siri, cannot read.

Signal is restricting screenshots in its Android and iOS apps so that AI agents built into phone operating systems, such as Gemini and Siri, cannot read on-screen messages. The company says these agents weaken its ability to guarantee privacy.

Key takeaways

  • Signal is limiting screenshots in its mobile apps on Android and iOS to stop operating-system AI agents from reading messages shown on screen.
  • The Verge reports that Signal’s leadership believes agentic AI features built into phones are undermining the app’s ability to guarantee privacy on its own terms.
  • End-to-end encryption protects messages in transit, but it cannot protect text once that text is displayed on a screen that other software can read.
  • Blocking screenshots also limits legitimate uses, such as saving a conversation for evidence or for personal records.
  • How well the restriction works will depend on what Android and iOS allow apps to control, and on how the AI agents capture screen content.

What is Signal changing in its mobile apps?

Signal, the encrypted messaging app run by a non-profit foundation, is restricting the ability to take screenshots inside its apps on Android and iOS devices. According to The Verge, the aim is to stop AI agents from reading users’ messages and other data shown in the app.

The agents in question are built into the operating systems themselves rather than installed as separate apps. Google’s Gemini on Android and Apple’s Siri on iOS are being developed to understand what is on a user’s screen, so they can answer questions about it, take actions, or learn more about the person using the phone. Signal’s concern is that this ability reaches into the content of private conversations.

Several details are not known from the available reporting. It is not clear whether the restriction is switched on by default for everyone or offered as an option, whether it works the same way on both platforms, or exactly which technical methods Signal uses on each.

Why is this in the news now?

The story is current because AI assistants are moving from answering typed questions to acting as agents. An agent may look at the screen, interpret what it sees and carry out tasks across apps for the user. Phone makers have been building these features into the core of their operating systems.

The Verge reports that Signal’s president described these on-device agentic capabilities as undermining Signal’s ability to independently guarantee privacy. The key word is “independently”. Signal’s model of privacy rests on the idea that the app alone controls who can read a message. When the operating system can look at the screen, a second party becomes involved, and Signal cannot fully control what that party does.

How do operating-system AI agents see what is on your screen?

A newcomer needs to understand one basic point. Encryption and screen visibility are separate problems.

Signal uses end-to-end encryption. Messages are scrambled on the sender’s device and only unscrambled on the recipient’s device, so the company’s servers and anyone intercepting traffic cannot read them. Once a message has been decrypted and shown on screen, though, it is ordinary text and images rendered by the phone.

The operating system draws everything on the display, so it can in principle capture it. AI features that offer to “look at your screen” usually work in one of two ways. They take a screenshot-like image and analyse it, or they read the structured information apps expose about their interface, such as accessibility data. The first method can often be blocked by an app. The second is harder to restrict, because accessibility tools exist to help people with disabilities and are built to read app content.

Android has long let apps mark windows as secure. Such windows are blank in screenshots and screen recordings. Signal has offered a screen security setting on Android for some time. iOS gives apps less direct control over screenshots, which is one reason it is not yet clear how Signal’s restriction works on Apple devices. Signal has taken a similar stance before on desktop, where it blocked screen capture in its Windows app in response to Microsoft’s Recall feature. Recall periodically records snapshots of the screen.

Who is affected, and how?

The most direct effect falls on Signal users with recent Android and iOS devices that have built-in AI assistants. For them, the change should make it harder for those assistants to analyse the contents of Signal chats. In practice, that means asking an assistant to summarise a conversation or act on something said in a chat may no longer work inside Signal.

Users also lose some ordinary convenience. Many people take screenshots of messages to share a funny exchange, keep a record of an address or arrangement, or save evidence of harassment. A screenshot restriction makes all of these harder. It does not stop someone photographing the phone with a second device, so it is a barrier against automated capture rather than a guarantee against copying.

The people a user talks to are affected as well. One reason Signal is acting is that the risk is not only to the person who owns the phone. If one participant’s operating system reads a conversation, the other participants’ words are exposed too, even though they never agreed to that AI feature.

Phone makers are affected in a different way. Their assistants become less capable inside one widely used privacy app. If other apps follow, the assistants could find whole categories of sensitive content off-limits.

Where do informed people disagree?

Privacy advocates broadly support the idea that encrypted apps should be able to keep their content away from system-level AI. They argue that people choose apps like Signal precisely because they want strong guarantees, and that an operating system reading the screen quietly removes those guarantees.

Others raise practical objections. Some users want their assistant to work everywhere, including in messaging apps, and see the restriction as Signal deciding on their behalf. Some say the choice should sit with the individual user, through a clear setting, rather than being imposed by the app or the platform.

There is also disagreement about where the real fix belongs. One view holds that apps blocking screenshots is a workaround. In this view, operating systems should give apps a reliable, standard way to declare content off-limits to AI agents, and should respect it. Another view says platform companies have commercial reasons to give their assistants broad access, so apps must protect themselves. How much processing happens on the device rather than in the cloud is also debated. Some argue that on-device analysis is less risky, while others note that data analysed locally can still be stored, logged or sent elsewhere.

What are the practical implications for users?

For most Signal users the visible change will probably be that screenshots inside the app fail or produce a blank image. Assistant features that rely on reading the screen may also stop working within Signal. Whether this can be turned off is not known from the reporting so far.

People who rely on screenshots for records may need other methods. These could include Signal’s own message-saving or export options where available, or simply writing down important details. Anyone keeping evidence of abuse should be aware of the change and consider how they will document conversations.

The change also offers a wider lesson. Encryption protects data in transit and on servers. It cannot protect text the device itself is allowed to read. Users who care about privacy may want to review which AI and accessibility features are switched on in their phone settings, and which apps have permission to see screen content.

What should people watch next?

Several questions remain open. The first is how Google and Apple respond, and whether they introduce formal ways for apps to exclude their content from AI agents. The second is whether Signal’s restriction holds up technically, particularly against agents that use accessibility data rather than screenshots.

It is also worth watching whether other messaging, banking and health apps take similar steps. If they do, phone assistants could face a patchwork of restricted apps, which may push platform companies towards a common standard. Finally, the balance between user choice and protection by default is likely to matter: whether Signal lets individuals opt back in, and whether regulators concerned with data protection take an interest in how system-level AI handles third-party content.

Frequently asked questions

Why is Signal blocking screenshots on my phone?

Signal is restricting screenshots in its Android and iOS apps so that AI agents built into phone operating systems cannot read your messages. The Verge reports that Signal believes these assistants, which analyse what is on screen, undermine its ability to guarantee privacy on its own. Blocking screen capture is a way to keep decrypted conversations away from software Signal does not control.

Does end-to-end encryption not already protect Signal messages?

End-to-end encryption protects messages while they travel between devices and while they sit on servers. Once a message is decrypted and shown on your screen, it is ordinary content the phone displays. The operating system, and AI features built into it, can potentially read anything on screen. Encryption alone cannot prevent that, which is why Signal is adding screen-level restrictions.

Can Gemini or Siri read my Signal chats?

Modern AI assistants on Android and iOS are designed to understand on-screen content so they can help with tasks. That ability could include text in messaging apps. Signal’s screenshot restriction is intended to stop this. How completely it works is not yet known. Assistants that rely on accessibility data rather than screen images may be harder for an app to block.

Can I still take screenshots in Signal?

Under the new restriction, screenshots inside Signal on Android and iOS are limited, so attempts may fail or produce a blank image. It has not been confirmed whether the restriction is on by default for everyone or whether users can switch it off in settings. Check the privacy settings in your version of the app to see which options are available.

Does blocking screenshots stop someone copying my messages?

No. A screenshot restriction blocks automated capture by the phone, including by AI agents, but it does not stop a person photographing the screen with another device. It also cannot stop the other person in a chat from writing down or repeating what you said. It reduces one specific risk rather than guaranteeing that a conversation can never be copied.

Will other apps block AI assistants in the same way?

It is not yet known. Apps handling sensitive information, such as banking, health or other messaging services, face similar concerns about system-level AI reading their content. Some already block screenshots for security reasons. Whether more follow Signal’s approach may depend on whether Google and Apple offer a standard way for apps to exclude content from their assistants.

Sources and further reading

  • The Verge: reporting on Signal restricting screenshots on mobile to block operating-system AI agents.
  • Signal Foundation: official blog posts and support pages on the app’s privacy and screen security settings.
  • Android developer documentation: guidance on secure windows and how apps can prevent screen capture.
  • Apple developer and platform security documentation: material on how iOS handles screenshots, screen recording and on-device intelligence.

Surfaced from the rss:verge signal “messaging app blocks AI”. AI-assisted draft, editorially reviewed.

Visited 1 times, 1 visit(s) today
share this recipe:
Facebook
X
WhatsApp
Telegram
Email
Reddit