Signal blocks screenshots by default on iOS 27 and Android to stop OS-level AI agents from reading on-screen content
Edited by Elliot Marsh, Editor · written with AI from the linked sources (how)
Signal now blocks screenshots in its mobile apps (iOS 27 and Android) by default, which it says should stop Siri, Gemini and other Android AI assistants from reading on-screen content. Users can turn screenshots back on in settings, with a warning that this may expose Signal content to AI agents. The protection is not available on macOS. Signal calls it incomplete because agents can still reach app data through the device filesystem or accessibility settings.
Entities: Signal, Meredith Whittaker, Apple, Siri, Google, Gemini
0 primary
What happened
Signal now blocks screenshots by default in its iOS 27 and Android apps. Signal says this should stop Siri, Gemini and other assistants built into the phone from reading what is on screen. Users can turn screenshots back on in settings and will see a warning that this may expose Signal content to AI agents. There is no equivalent protection on macOS, and Signal itself calls the fix incomplete because agents can still reach app data through the device filesystem or accessibility settings. The article gives no rollout date or version number for the change.
Why it matters
Signal users on recent iPhones and Android phones get a default that closes one route by which OS-level AI could read private conversations, and the fix costs them the ability to take screenshots. The more lasting issue is that app makers have few official controls over what system AI agents can see, so Signal is leaning on a workaround and asking Apple, Google and Microsoft for proper tools. Other privacy-focused apps may copy the approach, and regulators may cite it in debates about AI access to app data. Real protection depends on whether the platforms actually honour the block, which this change does not show.
What is noise
The headline overstates it: Signal is blocking one route (screen capture), not "Gemini and Siri" as a whole, and Signal admits the filesystem and accessibility routes remain open. "Existential threat to privacy at the application layer" is advocacy from Signal's president, not a measured finding, and no evidence is given that these agents currently read Signal content. The move also echoes Signal's earlier block of Microsoft Recall, so the novelty is modest.
Watch next
- 01Whether Apple and Google respond with documented developer controls for restricting AI agent access to app content, such as an official opt-out API, and when.
- 02Independent testing on iOS 27 and Android showing whether Siri, Gemini or accessibility-based agents can still read Signal content despite the screenshot block.
- 03Whether other messaging and privacy apps (WhatsApp, Telegram, banking apps) ship similar default blocks, and whether Signal reports how many users switch screenshots back on.
Coverage
2 storiesNewsletter
Get the no-hype AI news straight to your inbox
Weekly. No spam, unsubscribe any time. Privacy
More distribution signals
Full feed →- NYT-led publishers file summary judgment brief citing internal OpenAI/Microsoft messages calling AI training "astonishing theft" and admitting chatbots substitute for journalism18 Sept 202686
- Deepseek releases V4.1-Flash, an open-source model that sharply cuts KV cache memory and input-processing compute for AI agents10 Sept 202682
- Mistral releases Mistral Large 4, a ~1T-parameter open-weight multimodal MoE model, with day-0 Hugging Face support29 Oct 202680
- Meta releases Muse Spark Open, an open-weight ~816B-parameter multimodal mixture-of-experts model, with Hugging Face support15 Oct 202680