· via TechCrunch
Signal blocks mobile screenshots by default, citing AI agents that can read screens
Signal is disabling screenshots in its Android and iOS apps by default, arguing that OS-level AI agents able to scan screen contents threaten the privacy its encryption is meant to guarantee.
Signal is restricting the ability to take screenshots inside its app on Android and iOS, a change the company frames as a defensive response to operating-system-level AI agents that can observe what is on a user's screen.
According to TechCrunch, Signal President Meredith Whittaker announced the change on Friday, describing agentic AI built into Windows, Android and Apple's platforms as an "existential threat" to the messaging app's privacy model.
What changed
Signal already blocked screenshots of its Windows desktop app by default. The new step extends equivalent restrictions to the company's mobile apps.
The restriction is not absolute: users who want to take screenshots again can turn the setting off themselves. Whittaker cautioned that doing so would, in her words, "break our privacy guarantees," which signals that the company views the default as part of its security posture rather than a convenience toggle.
Why Signal is acting now
The core of Whittaker's argument, as reported by TechCrunch, is that agentic AI features being added to major operating systems can "scan" the contents of a user's screen. End-to-end encryption protects messages in transit and at rest inside the app, but anything rendered on screen — message text, contact names, conversation history — is visible to anything else that can read the display.
"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," Whittaker wrote. "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."
In other words, an AI assistant with screen access can effectively sidestep encryption without breaking any cryptography, because the plaintext is already on the display.
An admitted half measure
Whittaker acknowledged that the restriction is "incomplete." Blocking screenshots does not close every route an operating-system agent might take to application data, and she pointed to what she described as other, more sneaky ways AI agents could potentially access app data, without detailing them.
That admission frames the change as damage limitation rather than a fix. Signal can harden its own app, but it cannot control what the operating system running beneath it does with what it sees.
Why it matters
This appears to be one of the first cases of a major messaging app changing core behavior in response to AI agents rather than traditional threats such as malware or lawful interception. It exposes a structural weakness in encrypted communication: as platform vendors integrate assistants that can read and act on screen content, the line between "encrypted in transit" and "readable on device" starts to blur.
The move also raises questions about where responsibility lies. Signal can degrade its own usability to preserve its guarantees, but users who re-enable screenshots — or who use apps without such protections — remain exposed to a surveillance channel the app itself cannot close. If OS-level agents become standard equipment on Windows, Android and iOS, other privacy-focused apps face the same trade-off, and meaningful protection may ultimately require new operating-system-level assurances about what agents are allowed to read.
- #signal
- #privacy
- #encryption
- #ai-agents
- #messaging