Tag: Microsoft Recall

  • Signal Blocks OS AI Agents From Capturing Chats (2026)

    Signal Blocks OS AI Agents From Capturing Chats (2026)

    Takeaway: On November 20, 2026, Signal president Meredith Whittaker wrote that Signal is restricting OS-level AI agents from capturing and storing Signal data. The foundation says it will block screenshots in the app on iOS 27, and on Android it will restrict access when it can detect an AI assistant trying to read the Signal screen. That follows last year’s screenshot blocking aimed at Microsoft’s Recall feature on Windows 11. People can turn the protection off in Signal’s settings, but Signal says doing so can expose their chats — and their contacts’ chats — and break the app’s privacy guarantees. The post is explicit that the remedy is incomplete, and that Signal cannot currently implement these protections on macOS.

    The details below come from that official Signal blog post and its footnotes. This article does not add operating-system versions, app build numbers, or setting paths that Signal did not state.

    Why Signal says this is necessary

    Whittaker argues that Apple, Google, and Microsoft are integrating agentic features deep into the operating systems that Signal has to run on. The post names Gemini Intelligence with Google Gemini, and Onscreen Awareness with Siri AI, as examples, and compares the problem to Microsoft Recall on Windows. In Signal’s account, these agents scan what is on the screen so they can answer questions and carry out tasks. Once that happens, Signal data can be captured and made available to whatever local database or remote server the OS-level agent is using — outside Signal’s control, and with consequences for everyone in the conversation, not only the person running the agent.

    Signal presents this as more than a settings dispute. The foundation says its privacy guarantee has long rested on an open-source codebase and a formally verified encryption protocol. OS-level screen capture, it argues, undermines that guarantee and is “an existential threat to Signal, and our ability to responsibly uphold our mission.” The same week’s Gemini computer use rollout on Mac and earlier Gemini chat-history and training controls are separate Google product stories; Signal’s post is about OS agents reading app content, not about those Gemini-app switches.

    What Signal is changing, and on which platforms

    The post describes three layers of protection, with different coverage on each platform:

    • Windows: Signal already used screenshot blocking to prevent Microsoft’s Recall AI from taking and storing screenshots of Signal messages on Windows 11 by default. A footnote says Recall captures the screen every few seconds, extracts text with optical character recognition, and stores it in a searchable on-device database, which could retain Signal data — including disappearing messages.
    • iOS 27: Signal is preventing screenshots in the app, with the same goal of stopping OS agents from capturing what is on the Signal screen.
    • Android: Signal says it will restrict agentic systems on versions where it can detect an AI assistant attempting to read the Signal screen. It automatically enables the “Screen security” setting in Signal Android when that kind of access is detected.

    The Android footnote is careful about fragmentation. There are many Android versions beyond Google’s Pixel builds, and for many vendors Signal says it does not have a reliable way to detect a screenshot attempt if screenshots are already disabled. Rather than make screenshots stop working in a way that would confuse people, Signal says it focused on detecting when an AI assistant tries to access the screen.

    What you can change, and what Signal cannot cover

    People who want to take screenshots can disable the protection in the Signal app’s settings. Signal is blunt about the trade-off: doing so can expose your content, and your contacts’ content, to these AI agents, and it breaks Signal’s privacy guarantees.

    The foundation also says the remedy is incomplete. Restricting screenshots does not stop other ways an agent might reach application data, such as files already on the device. Signal adds that it currently cannot implement these protections on macOS. The reason, in footnote 3, is that Signal cannot reliably block screenshots on macOS, and that existing continuous-capture protections do not apply to the screenshot path most agentic systems use. The post says Signal is also not blocking Siri AI on macOS. For macOS 27, the only recommendation in the footnote is to add Signal to the Excluded Apps section in System Settings > Siri, to protect message content from proactive access by Siri AI.

    Signal also says it is struggling with a lack of documentation about how these agents actually work. The post says engineers have spent recent months on forensic investigations across Android and iOS to learn what happens when the systems access application data.

    What Signal wants from Apple, Google, and Microsoft

    The foundation calls on the major consumer OS companies to give developers simple technical tools and accurate documentation so apps can keep AI agents away from sensitive data. It asks for two remedies:

    • An easy, well-documented way for a developer to opt an app out of AI data access and processing, including screen reading, filesystem access, notifications, APIs, and other OS frameworks, without undermining the app’s core utility.
    • Better documentation and technical specifications for how these agents function, especially around data access and processing. Signal says confidential computing approaches are not a complete solution, because they still leave the extent of any privacy guarantee in the provider’s hands.

    The post also flags third-party agents, naming Meta’s Muse as an example, that may reach application data through deep permissions — including accessibility settings that were meant for people, not as a “skeleton key” for AI agents. That is a broader platform concern; it is not a report about the official Muse Chrome extension Meta announced this week.

    Signal says it is talking with others in the industry and is “cautiously optimistic” that some major OS providers, including Google and Apple, are engaging on how these systems should work with privacy-sensitive apps. It hopes to have more to share, and it frames itself as a canary: the risk, it argues, is not limited to Signal.

    Key facts

    ItemWhat official sources say
    AnnouncementSignal blog post by Meredith Whittaker, dated 20 November 2026.
    What is changingSignal is restricting OS-level AI agents from capturing and storing Signal data.
    WindowsScreenshot blocking was already used against Microsoft Recall on Windows 11 by default.
    iOS 27Signal is preventing screenshots in the app.
    AndroidSignal enables the “Screen security” setting when it detects an AI assistant attempting to access the screen, on versions where it can detect that access.
    User overrideThe protection can be disabled in Signal’s settings, with the official warning that this can expose chats and break Signal’s privacy guarantees.
    macOSSignal says it cannot currently implement these protections. For macOS 27, it recommends adding Signal to Excluded Apps in System Settings > Siri.
    LimitsScreenshot blocking does not cover other paths such as filesystem access. Android coverage depends on Signal’s ability to detect the agent. Signal also asks OS vendors for a proper opt-out and better documentation.

    FAQ

    What did Signal announce on November 20, 2026?

    Signal said it is restricting the latest AI agents built into mobile operating systems from potentially capturing and storing Signal data. The post is signed by Meredith Whittaker and is on the official Signal blog.

    Does this apply to iPhone and Android?

    On iOS 27, Signal is preventing screenshots in the app. On Android, it restricts access when it can detect an AI assistant trying to read the Signal screen, and it automatically turns on the “Screen security” setting in that case. Signal says it cannot reliably detect this on every Android vendor’s software.

    What about Windows and Microsoft Recall?

    Signal says it already used screenshot blocking last year to prevent Microsoft’s Recall AI from taking and storing screenshots of Signal messages on Windows 11 by default. The new post extends the same concern to agentic features on iOS 27 and Android.

    Can I still take screenshots in Signal?

    Yes, if you disable the protection in the Signal app’s settings. Signal warns that doing so can expose your content and your contacts’ content to OS AI agents, and that it breaks the app’s privacy guarantees.

    Is macOS covered?

    No. Signal says it is currently unable to implement these protections on macOS, because it cannot reliably block screenshots there, and it is not blocking Siri AI on macOS. For macOS 27, the post recommends adding Signal to the Excluded Apps section in System Settings > Siri.

    Why does Signal say screenshot blocking is not enough?

    The post says restricting screenshots does not address other ways AI agents could reach application data, such as content already on the device via the filesystem, and that developers lack a clean way to opt out of screen reading, filesystem access, notifications, and other OS frameworks.

    Is this a Signal app version number I should look for?

    The November 20 blog post does not name a Signal app version or a rollout date by store build. It describes the policy and the platform-specific measures. This article does not invent a version number.

    What is Signal asking Apple, Google, and Microsoft to do?

    It wants an easy, documented way for developers to opt an app out of AI data access — including screen reading, filesystem access, notifications, APIs, and other OS frameworks — without breaking the app, plus clearer documentation of how these agents access and process data.