Signal says it blocks iOS 27 screenshots and detects AI assistants on some Android versions, but not on macOS
Signal's post of 20 November 2026 gives four platforms four different answers to AI assistants that read the screen, 548 days after its Windows answer to Recall (derived). It says its fix is incomplete; no saved page shows an assistant reading a Signal chat, or a reply from Apple or Google.
By Parminder Kumar Sharma · LinkedIn · · 18 min read
Four platforms, four different answers, 548 days after the Windows one (derived)
Signal's post of 20 November 2026, Signal is Serious About Protecting Your Privacy, addresses four platforms and gives each a different answer to the AI assistants built into the operating system. On iOS 27 Signal says it is preventing screenshots in the app. On Android it says it switches on its "Screen security" setting automatically when it detects an AI assistant attempting to read the screen, on "Android versions where we can detect". On macOS it says it is not blocking Siri AI and cannot reliably block screenshots, and it gives users one piece of advice. On Windows it points back to its earlier answer to Microsoft Recall. That earlier post, By Default, Signal Doesn't Recall, is dated 21 May 2025 and put the same setting on by default in Signal Desktop on Windows 11. From 21 May 2025 to 20 November 2026 is 548 days (derived). BleepingComputer's relay is dated 22 November 2026, 10:15 AM as the page prints it, 2 days after the post (derived). This site saved the pages between about 16:04 and 16:07 UTC on 22 November 2026.
What this does not establish. The post describes a design decision and a dispute with the makers of operating systems. It reports no incident: it does not say that Siri, Gemini or any other assistant has read a Signal conversation in the field. It says Signal spent "considerable developer time and energy doing forensic investigations across Android and iOS" to learn what the assistants do. It prints no Signal version number, no list of Android versions or vendors and no label for the iOS setting, and it says its own remedy is incomplete. No saved page carries a reply from Apple or Google to Signal's description. This site read no Apple page, no Signal app or support page, and tested none of the protections.
Whose interest. The post describes Signal Technology Foundation as the nonprofit home of the Signal app and a US 501(c)(3) tax-exempt organization. It is a party to the dispute it describes, and it calls the situation "an existential threat to Signal". Google and Apple make the operating systems and the assistants in question. The Google pages this site saved are Google's own announcements of its assistant, and no Apple page was saved. EFF is an advocacy nonprofit; its October post says the setting it reports fulfils one of the asks of its Encrypt It Already campaign. BleepingComputer is a news publisher, and its saved page carries an advertisement for an unrelated email security product. No vendor pays this site, and it shares no vendor's interest.
Platform by platform: what the post says Signal does
Diagram 1 sets the four answers out above a to-scale axis of the dated pages. The table adds what the post leaves open for each platform. Everything in the middle column is Signal's own statement; this site tested none of it.
What Signal's post of 20 November 2026 says for each platform, and what it leaves open (Signal's 2025 post for Windows; saved 22 November 2026, 16:05 and 16:07 UTC)
| Platform | What Signal says it does | What the page leaves open |
|---|---|---|
| iOS 27 | Prevents screenshots in the app. Users can turn the protection off in the app's settings; Signal says that may expose their and their contacts' content to the AI agents. | Which Signal versions carry it. The label of the setting (not printed). Whether it stops Siri's onscreen awareness in practice (Signal's claim; untested here). |
| Android | Switches on "Screen security" automatically when it detects an AI assistant attempting to access the screen, on "Android versions where we can detect". | Which Android versions and vendors. How detection works. Whether notification content is covered. |
| macOS | Does not block. Signal says it cannot reliably block screenshots on macOS. Its only advice: add Signal to Excluded Apps in System Settings > Siri, to protect message content from proactive access by Siri AI. | Whether the Excluded Apps setting covers anything beyond proactive access; the post does not say. |
| Windows | In 2025, Screen security on by default in Signal Desktop on Windows 11, to keep Microsoft Recall from capturing screenshots (post of 21 May 2025). | The 2026 post says only that Signal did this "last year"; it does not restate the Windows position or Recall's current state. |
One label, two mechanisms, one described. The setting is called "Screen security" in the 2025 post, where it belongs to Signal Desktop on Windows, and in the 2026 footnote about Android. The 2025 post says how it works on Windows: Signal sets a Digital Rights Management flag on the application window, which, in the post's quotation of Microsoft's developer documentation, means "content won't show up in Recall or any other screenshot application". The 2026 post does not say how the Android setting works or how Signal detects assistant access. The same name, and the saved pages explain the mechanism only for Windows.
Why Android is detection, not a block. Footnote 2 says: "For many vendors, we don't have a reliable way to detect when a screenshot was attempted if screenshots are already disabled." Signal says it chose not to make screenshots simply stop working, which it describes as confusing users, and to focus on detecting an assistant's attempt to access the screen instead. That is why coverage depends on detection, and why the post limits itself to Android versions where Signal can detect.
Windows then, macOS now. In 2025 Signal wrote that Microsoft had launched Recall "without granular settings for app developers", while, in the post's account, Microsoft's developer documentation described the flag above. In 2026, for macOS, it writes of "a lack of clearly documented APIs", says protections against continuous screen capture exist but cannot be applied to screenshots, "which is how most agentic systems access data from applications like Signal", and says some APIs "have been deprecated in favor of screen-protection solutions that only work for streaming media and other DRM-protected content". The post gives a second reason in the body: Google, Apple, Microsoft and others do not let developers restrict AI data access "without making drastic compromises to the core user experience", and "This is also the reason we are currently unable to implement these protections on macOS." These are Signal's descriptions. This site checked no Apple or Microsoft documentation.
Where BleepingComputer's relay and the post differ
BleepingComputer's article relays the post and adds lines of its own. The table sets them side by side. Where the relay prints something the post does not, this site says so and does not decide which is right: it has not seen Signal's app or a Signal support page.
BleepingComputer, 22 November 2026, against Signal's post of 20 November 2026 (both saved 22 November 2026; quotations copied from the pages)
| Point | BleepingComputer's relay | Signal's post |
|---|---|---|
| How far the block reaches | Signal's engineers designed protections "to block all extraction pathways verified through reverse engineering". Later it adds that AI agents could still reach conversations by other methods. | "Our remedy is also incomplete", and filesystem retrieval of content already on the device is named as a route it does not address. It speaks of forensic investigations; the words reverse engineering and all pathways are not in it. |
| The iOS setting | Go to Settings > App Security and check that "Block Siri AI" is enabled. An image is captioned "Option to block Siri scanning on Signal for iOS"; its alt text reads "Option to disable Siri screenshots on iOS". This site saved the page text, not the image. | Users can turn the protection off "within the Signal app's settings". Neither the path nor the label is printed. |
| Updating | Signal "recommends that users upgrade to the latest available version". | The post does not mention updating or upgrading. |
| Automatic or not | Protections "are automatically activated when possible"; no app version numbers were given. | Android: switched on automatically when assistant access is detected. iOS: the post says Signal is preventing screenshots and that users can disable it, and does not say whether it is on by default. |
| Where they agree | No version numbers given. macOS: Signal is "currently unable to provide any screenshot-blocking protection"; for macOS 27 it recommends Excluded Apps "to prevent Siri AI from proactively accessing message content". | No version numbers printed. macOS: not blocked; Excluded Apps advised "to protect message content from proactive access by Siri AI". |
The relay attributes the post to Signal's president. The post itself carries a byline, and this site names nobody. The relay's headline says Signal will "block Siri, Gemini AI from capturing chats". The post says Signal is "restricting" the agents and "blocking such capture", by preventing screenshots on iOS 27 and, on Android, when assistants attempt to capture screen data; the Android measure depends on detection. On iOS the relay adds that the protection "also prevents users from taking screenshots of their Signal conversations, even for legitimate purposes"; the post says that people who want to take screenshots can disable the protection.
EFF's two posts on the same iOS path
EFF's two posts are dated by their own page metadata, not by printed text. The first, How to limit what Apple's new Siri AI can access in iOS 27, carries a published stamp of 18 September 2026 (63 days before Signal's post, derived) and a modified stamp of 28 October 2026. The second, Victory: you can now exclude apps from Siri AI, carries a published stamp of 28 October 2026, 2 hours 2 minutes 53 seconds after the first page's modified stamp (derived from the two stamps, both in the same offset) and 23 days before Signal's post (derived). This site read no Apple page, so what follows is EFF's account of Apple's software, with EFF's dates.
18 September, as saved. EFF says that by default Siri AI searches Apple's own apps, and that if a third party app developer does not add support, Siri AI "won't be able to access the contents of that app", unless it is shared screenshot style through a feature called on-screen awareness. It then says there is currently no way for users or app developers to block that feature, and that a means for developers to block it would be "a large improvement to privacy". The page also says it looks as if Apple is planning a single privacy toggle in a future update.
28 October. EFF reports that iOS 27.2 added a setting to exclude apps from Siri AI: Settings > Siri > Excluded Apps. Its words for the result: "Now Siri AI cannot access the contents of the app." Its example is Signal: "if you exclude Signal, you cannot invoke Siri and ask it to discern any information in the Signal app", and even on-screen awareness no longer works for an excluded app. EFF adds that at launch it found Siri AI's app access settings pretty confusing.
The difference, left as a difference. EFF's September page, as saved after its own 28 October edit, still says there is currently no way to block on-screen awareness. EFF's October post, published two hours later, says an excluded app is out of reach of on-screen awareness. Signal's post, 23 days after the October one, advises Excluded Apps for macOS 27 only and does not mention it for iOS, where it describes a screenshot block. Signal also says, in footnote 3, that screenshots are how most agentic systems access data from apps like Signal. Whether Signal's block and the Excluded Apps setting overlap on iOS 27.2, or whether either one alone stops Siri's on-screen awareness from reading Signal, is stated by none of the three pages, and this site tested nothing.
Android: Google's two posts and EFF's account
Google's two posts, A smarter, more proactive Android with Gemini Intelligence and Android's Agentic Future, are both dated 12 May 2026, 192 days before Signal's post (derived). They describe user controls. Connecting Gemini to Autofill with Google "is strictly opt-in"; Gemini "can only access the apps you allow it to work in, and not the rest of your device"; while Gemini automates an app a notification chip appears that you "can't dismiss"; and the page says Google will "soon" extend the Android Privacy Dashboard to show which AI assistants were active and which apps they used in the last 24 hours. It also says users will be able to turn on Gemini app automation for specific apps in settings "later this year". The first post says features roll out first on the latest Samsung Galaxy and Google Pixel phones, and adds screen or image context to app automation: long press the power button over a list and ask Gemini to act on it.
What the two Google pages do not say. Neither page names a control for an app developer to opt out of screen context. The only places the second page mentions developers are its stated goal for Android to be "an open platform where third-party developers and device manufacturers can build unique and trusted agentic experiences", and its sentence that Google is working with the developer ecosystem to promote its security and privacy practices "as industry standards for all AI assistants running on our platform". The first page does not use the word developer at all. This site read two pages and not Android's developer documentation.
EFF's account, 28 October. EFF says Gemini's controls have improved a little but lack "the ability to set the screen context feature per-app". It says Signal does not appear in Gemini's Connected Apps list because Signal "doesn't support Gemini's app access feature at all", so Gemini cannot read Signal's content directly. But, EFF says, Gemini can read notifications if the Device Assistance Connected App is enabled, which EFF says you likely have, and can use screen context unless the user has turned off the "Use screen and app data" option. "These two features may access the contents of the Signal app or notifications, if you choose to use them." EFF asks Google to let users turn these features off per app.
Put beside Signal's Android measure. Signal says its Android protection is triggered when an AI assistant attempts to access the screen. EFF's October post describes two routes, notifications and screen context. Signal's post does not say whether its detection covers the notification route, and no saved page says whether it catches the screen context route on any particular Android version or vendor. Signal's list of things it wants developers to be able to opt out of includes notifications, which states a request; the post does not say whether the current protection touches notifications.
Signal's stated stake, and what it asks for
Signal's argument, in its own framing: Google, Apple and Microsoft make the three major consumer operating systems, which it calls "the oceans in which Signal, and other applications, must swim". The assistants read what is on screen, capture Signal data, and make it available to "whatever local database or remote server an OS-level AI agent is using at that moment", which Signal says undermines its ability to guarantee privacy independently. A footnote describes Recall: it "captures your screen every few seconds, extracts text from screenshots via optical character recognition, then populates a searchable on-device database".
The two asks. First, that developers have an easy and well documented way to opt an app out of AI data access and processing, "including (but not limited to) screen reading, filesystem access, notifications, APIs, and other OS frameworks", without undermining the app's core utility. Second, that operating system companies document how the AI agents work, particularly their data access and processing. Signal adds that "Confidential computing approaches are not a complete solution", because they "still leave the power to determine the extent of privacy guarantees (if any) in the hands of these providers".
Third party agents. The post also calls for tools that stop third party agents, "like Meta's Muse", from reaching app data with the deep permissions users are encouraged to grant them at setup, and says many such agents use accessibility settings meant for people. This site's separate brief on NebuSec's lab demonstration against Meta Muse covers a different subject and says nothing about Signal.
The vendors. Signal says it is "cautiously optimistic that some major OS providers, like Google and Apple, are engaging with us". The sentence does not name Microsoft. No saved page carries a response from Apple, Google or Microsoft.
What the pages state, and what they do not
Stated and not stated across the saved pages (saved 22 November 2026, about 16:04 to 16:07 UTC)
| Question | What the saved pages state | What they do not state |
|---|---|---|
| Which platforms? | iOS 27, Android, macOS, Windows: four treatments (counted by this site) | Signal versions with the protections; the iOS setting's label |
| Which Android devices? | "Android versions where we can detect" | Which versions and vendors; how detection works |
| Has an assistant read a Signal chat? | Nothing: the post describes forensic investigation | Any field case, from either assistant |
| Is the fix complete? | Signal: incomplete; filesystem named | Any other route, and how it would be closed |
| What do Apple and Google say? | Signal: cautiously optimistic that both are engaging | Any statement from either; any Apple page (none saved) |
| Is there a developer opt out? | Google's two posts name user controls; EFF: no per app screen context setting in Gemini | A developer opt out on Android, or Apple's documentation of one |
| Do the iOS protections overlap? | EFF: Excluded Apps stops Siri reading an excluded app on 27.2. Signal: screenshots blocked | Whether they overlap or either suffices; untested here |
| How many people, and who else? | Nothing on either | Users on iOS 27 or covered Android builds; other messaging apps' choices |
What to do, in order
First the primary's advice, then the relay's, then this site's suggestions, each marked.
Take this with you
Actions, in the order worth doing
- Signal's advice, for macOS 27: add Signal to Excluded Apps in System Settings > Siri, to protect message content from proactive access by Siri AI.
- Signal's note, for everyone: people who want screenshots can turn the protection off in Signal's settings, and doing so may expose their and their contacts' content to the AI agents.
- BleepingComputer's relay, not in Signal's post: upgrade to the latest Signal and check Settings > App Security for Block Siri AI on iOS. This site has not seen the app, so treat the path as unconfirmed.
- This site's suggestion, iPhone: per EFF's October post, open Settings > Siri > Excluded Apps (iOS 27.2) and review which apps you do not want Siri AI to read. Whether that overlaps with Signal's screenshot block is not stated.
- This site's suggestion, Android: per EFF, review Gemini's Connected Apps, whether Device Assistance is on, and the Use screen and app data option. EFF reports no per app control.
- This site's suggestion, organisations: add OS level assistants to the data egress model for messaging, and record which apps can be excluded and which cannot. Signal gives no versions, so record the Signal and OS versions in use.
- This site's suggestion: ask each OS vendor in writing for the documented developer opt out Signal requests, and what its documentation says about screen, file and notification access.
- This site's suggestion: review which third party agents hold accessibility permissions on managed devices, since Signal's post says such agents use them.
- This site's suggestion: do not treat a screenshot block as a complete control. Signal itself says it is not.
The settings the pages name
None of the saved pages prints code, a version number, a flag or an indicator. They do print setting names and paths. Each line below is attributed to the page that prints it, with its date, and the layout is this site's. One line, the iOS toggle, is printed only by the relay and not by Signal's post. A defender can use the list as a checklist of places to look on a test device; this site has not opened any of them.
Signal post, 20 Nov 2026, macOS 27: System Settings > Siri > Excluded Apps
Signal post, 20 Nov 2026, Android: the "Screen security" setting in Signal Android, enabled automatically when assistant access is detected
Signal post, 21 May 2025, Windows 11: Signal Settings → Privacy → Screen security
BleepingComputer, 22 Nov 2026, iOS: Settings > App Security > Block Siri AI (relay only; not in Signal's post)
EFF, 28 Oct 2026, iOS 27.2: Settings > Siri > Excluded Apps > Add Apps
EFF, 28 Oct 2026, Gemini app: Settings > Personal Intelligence > Connected Apps
EFF, 28 Oct 2026, Gemini settings: Device controls > Screen context > Use screen and app data
The question the post leaves open
Signal has told its users what it blocks, where it cannot, and what it wants from Apple, Google and Microsoft. It has not said whether any assistant has read a conversation, which builds are covered, or what the OS vendors reply. So: which assistant on your devices can read the screen of the apps you treat as private, and who documented it?
Key facts
Sources
- PrimarySignal, Signal is Serious About Protecting Your Privacy, dated 20 November 2026, saved at about 16:05 UTC on 22 November 2026: what Signal restricts on iOS 27, Android and macOS, the three footnotes, the two asksSignal (blog)accessed 2026-11-22
- PrimarySignal, By Default, Signal Doesn't Recall, dated 21 May 2025, saved at about 16:07 UTC on 22 November 2026: the Windows 11 Screen security default and the Recall precedentSignal (blog)accessed 2026-11-22
- PrimaryGoogle, A smarter, more proactive Android with Gemini Intelligence, 12 May 2026, saved at about 16:07 UTC on 22 November 2026: opt in controls, screen context and rolloutGoogle (blog)accessed 2026-11-22
- PrimaryGoogle, Android's Agentic Future: Building Gemini Intelligence on a Foundation of Security and Privacy, 12 May 2026, saved at about 16:07 UTC on 22 November 2026: user controls and the developer goalGoogle (blog)accessed 2026-11-22
- Reported byBleepingComputer, Signal to block Siri, Gemini AI from capturing chats on iOS, Android, 22 November 2026 10:15 AM as printed, saved at about 16:04 UTC: a relay of the post with lines the post does not carryBleepingComputeraccessed 2026-11-22
- Reported byEFF, How to limit what Apple's new Siri AI can access in iOS 27, published 18 September 2026 and modified 28 October 2026 per its metadata, saved at about 16:07 UTC on 22 November 2026: an advocacy nonprofit's account of Siri AI settingsElectronic Frontier Foundationaccessed 2026-11-22
- Reported byEFF, Victory: you can now exclude apps from Siri AI, published 28 October 2026 per its metadata, saved at about 16:07 UTC on 22 November 2026: Excluded Apps on iOS 27.2 and Gemini controlsElectronic Frontier Foundationaccessed 2026-11-22


