Signal blocks in-app screenshots, calls AI agents ‘existential threat’ to privacy mission

Wait 5 sec.

As tech giants look to embed personal AI agents in users’ everyday digital interactions, Signal has said it is taking steps to address the new privacy risks they pose, ensuring that users’ conversations on its secure messaging app remain out of the agents’ reach.The end-to-end encrypted platform has announced that it will no longer allow users to take screenshots within the Signal app by default. The move is aimed at “restricting the latest AI agents built into mobile operating systems from potentially capturing and storing your Signal data,” Meredith Whittaker, the president of the non-profit entity behind the app, wrote in a blog post published on Friday, November 20.While Signal has already had screenshot blocks in place for its Windows app users, the latest announcement brings similar restrictions to Android and iOS users as well. Users, however, can disable the block from within the app’s settings.It underscores one of the key tensions in the AI era, with users having to weigh the convenience of AI agents against the privacy risks of giving them access to personal information. Many have welcomed AI agents into their digital lives, using them for a wide range of tasks such as booking dinner reservations, planning trips, organising email inboxes, handling shopping purchases, and much more. But these autonomous agents are only as useful as the personal information and user context shared with them.The more they know about users, the better they perform tasks on their behalf, potentially gaining insights not just into an individual’s life but also that of their friends, family, and colleagues. This, in turn, has raised new privacy concerns, with experts warning that users may not fully anticipate the risks when handing over such large amounts of their personal data.Also Read | After WhatsApp, India sends notices to Telegram and Signal over anonymity featuresDescribing it as an “existential threat” to the app, Whittaker argued that giving AI agents the ability to “scan” the information on users’ screens, Windows, Android, and macOS/iOS “are undermining Signal’s ability to independently guarantee privacy.”“Now, in service of AI agentic capabilities, your sensitive data is being captured, stored, and made accessible to whatever local database or remote server an OS-level AI agent is using at that moment. All of this is outside of Signal’s control once the data has been captured, and imperils the privacy of all the participants in a conversation,” she said.Story continues below this adHowever, Whittaker also acknowledged that screenshot blocks were an “incomplete” solution, as it does not “address other more sneaky ways AI agents could potentially access application data.”In 2025, Signal rolled out similar restrictions to block Microsoft’s Recall AI feature from taking and storing screenshots of users’ Signal messages on Windows 11 operating systems by default.Among major operating system providers, Google has been integrating agentic capabilities such as Gemini Intelligence, which can read the contents of a user’s screen and take actions. Apple also rolled out similar on-screen awareness capabilities as part of its AI-revamped Siri voice assistant launched earlier this year, along with the iOS 27 update.“The net result is that these AI capabilities in the operating system are undermining Signal’s ability to independently guarantee privacy,” Whittaker wrote.Story continues below this adSignal is widely considered to be a leading easy-to-use, encrypted messaging service among cybersecurity experts, since there are no public reports of hackers intercepting users’ messages in-transit by cracking the platform’s end-to-end encryption protocol. However, Signal is only as secure as the security of a device. Encrypted messages can be leaked if an attacker gains access to an unlocked device, installs spyware, or tricks a user into linking their account to a malicious device.Also Read | Google DeepMind used AI to test Signal’s encryption: Here’s what it foundMost computer-use agents work by taking repeated screenshots in a loop to see what is on the screen, decide on an action, and check the result.Whittaker said that OS-level AI agents capture users’ sensitive data and store it in a local database or remote server being used at that moment. “All of this is outside of Signal’s control once the data has been captured, and imperils the privacy of all the participants in a conversation,” she added.For iOS 27 devices, Signal automatically blocks Siri AI from taking screenshots in the app unless disabled. In the case of Android users, the platform first detects whether AI agents are attempting to capture screen data and blocks them. The platform said it is currently unable to implement screenshot blocks on macOS due to lack of developer access by Apple.Story continues below this adOpenAI, Anthropic, and Meta have also introduced their own versions of personal AI agents. Whittaker said that there is a growing need to prevent agents like Meta’s Muse from “accessing application data using the deep permissions that users are encouraged to grant them during setup.”She further called on major OS vendors such as Apple, Google, and Microsoft to “provide simple technical tools and accurate documentation that give [app] developers like Signal ways to prevent AI agents from accessing sensitive application data.”