Signal Restricts Screenshots to Block AI Agents From Reading Private Chats

Signal announced on Friday that it will restrict screenshots and block AI assistants from reading private chats on mobile phones, according to Forbes Mexico.

The messaging app warned that the tech industry’s rush to embed artificial intelligence into consumer devices is actively undermining its core security model. Considered a benchmark for secure messaging by privacy advocates, the organization described its new defensive safeguards as incomplete.

Operating system features designed to observe on-screen activity pose an existential threat to end-to-end encryption, according to a corporate blog post shared with Reuters before its public release, as cited by Forbes Mexico.

“Currently, these AI integrations are transforming said environment, making it increasingly hostile for privacy. This represents an existential threat to Signal and to our capacity to fulfill our mission in a responsible manner,” the company said.

Operating system threats

The platform pointed to built-in system capabilities like Apple’s Siri and Google’s Gemini as particularly concerning due to their deep integration with user hardware, according to Forbes Mexico.

In an update detailed by AI.info, Meredith Whittaker stated on November 20 that these protections aim to stop operating system agents from capturing and storing message data. Whittaker also said that Signal has spent considerable development time researching how agents access app data on Android and iOS due to a lack of documentation.

Signal Restricts Screenshots to Block AI Agents From Reading Private Chats
Photo: ai.info

These monitoring tools target functions such as Gemini Intelligence and Siri AI Onscreen Awareness, alongside third-party agents like Meta’s Muse, which the company argues might exploit granted permissions, as reported by AI.info. Signal said that an AI agent could still access application data through alternative methods, such as retrieving content already stored on a device via the file system.

Signal previously implemented a similar block to stop Microsoft’s Recall feature from taking automated snapshots of user screens on Windows, noting that the feature captures the screen every few seconds, extracts the text, and stores the information in a local searchable database, according to Tiempo. However, the company acknowledged it cannot deploy reliable screen protections on Apple’s macOS.

Device controls and mobile settings

Under the newly deployed configuration, the application blocks screen captures by default. On iOS, the software introduces a specific toggle labeled Block Siri AI, while Android devices automatically engage screen security protocols upon detecting an AI assistant attempt to access the display, according to AI.info.

Signal said that for many Android manufacturers, it lacks a reliable method to detect a screenshot attempt when screenshots are already disabled, leading the company to focus on detecting AI assistant access.

Pop-up warnings will appear whenever the software detects capture attempts or assistant interactions, allowing users to modify their device configurations. Whittaker warned that disabling these protections exposes user chats and contact data to system agents.

For macOS users, where reliable capture blocking remains technically unfeasible, the company recommends adding the messaging client to the Excluded Apps section within Siri system settings.

Signal capturas pantalla IA
Photo: que.es

Developer demands and system changes

The messaging provider formally requested that Apple, Google, and Microsoft provide concrete developer tools to exclude applications from AI data processing, according to Forbes. The organization also urged these tech giants to deliver greater transparency regarding the exact data harvested and analyzed by their algorithms.

Before leading the organization, Meredith Whittaker spent 13 years at Google, where she founded the company’s Open Research group and co-founded M-Lab. In 2018, she was a core organizer of the Google Walkouts, leading thousands of employees in protesting the firm’s culture and its involvement in military technology before resigning in 2019.

Her career also includes serving as a senior advisor on AI to Chair Lina Khan at the Federal Trade Commission, holding the role of Minderoo Research Professor at New York University, and co-founding the AI Now Institute. In 2023, TIME magazine recognized her among the 100 most influential people in artificial intelligence.

The company said that operating systems represent the fundamental environment where secure apps must function, yet current AI additions turn that environment hostile. The messaging provider maintains that third-party developers require granular controls over screen reading, file system access, and system frameworks to properly protect private communications.

Photo of author

Technology Desk

Technology Desk at Archyde.

CAS Mandates Russia Return; Canada Said to Refuse Visas, IIHF Reportedly Hacked