Don’t ask your AI agent to read and decipher that lengthy text message. And don’t screenshot it either. It may seem harmless, but Signal’s latest setting underlines that it isn’t. The privacy-focused messaging and calling app is rolling out a new default setting to stop built-in smartphone AI agents from potentially seeing and saving your Signal data. In a blog post announcing the change, Signal President Meredith Whittaker said these AI bots scan what’s on your screen to get to know you, capturing your data and making it available to AI systems to fulfill your requests, such as answering questions you may ask. That threatens Signal’s mission to guarantee privacy, Whittaker wrote. So it’s putting a stop to it with a default setting to block screenshots in the Signal app on iOS 27 and when AI assistants try to capture information on Android screens by default. Signal’s not new to blocking AI bots Whittaker said that this isn’t the first time Signal has taken such measures against an operating system. Last year, it blocked screenshots of Signal messages from Microsoft’s Recall AI on Windows 11. Now the company is taking a similar approach to Apple’s iOS 27 and Android operating systems, where AI assistants try to read your Signal screen. The problem is the same for the three major OS companies. Microsoft, Google and Apple are incorporating AI into their operating systems, and Signal says that compromises the space for third-party developers, including Signal, to build apps for these systems. “Right now, these agentic AI integrations are changing this environment in a way that’s making it increasingly hostile to privacy,” Whittaker wrote. Blocking screenshots isn’t the be-all and end-all Whittaker said the screenshot setting won’t fix everything, because AI agents can still sneakily access data on your phone. Tech companies should take steps to give developers more choice, such as an opt-out to disable AI agent access, and provide more documentation on how big tech AI agents work. “Even where we are able to detect and block AI data access, Google, Apple, Microsoft and others do not provide ways for developers to easily restrict such AI data access without making drastic compromises to the core user experience,” Whittaker said. “This is also the reason we are currently unable to implement these protections on MacOS.” A way around encryption In most cases, text messages are encrypted when they travel between devices. But AI agents can bypass that. John Strand, CEO of Black Hills Information Security, explained that when you access the information, it has to be decrypted because you cannot read encrypted data. “So, if an AI agent can access screenshots or directly observe what’s happening on your screen, it can read those messages without ever having to break the encryption itself,” said Strand. “That’s the scary part. You can have incredibly strong encryption, but if something can see what the user sees, that encryption isn’t going to protect you at that point.” You can still take screenshots, but beware If you want to take screenshots in Signal, you can disable the setting, but that could mean AI agents could access you and your senders’ information, like text messages and shared files such as photos. Strand said the worst-case scenario is that an attacker could see everything you see, from private conversations to passwords. And that’s what Signal is trying to avoid. “They want to make sure that if something like this happens, they’ve done absolutely everything they reasonably can to protect their customers and the information being accessed through their application,” Strand said. Signal, Apple and Google did not immediately respond to requests for comment.
Signal Is Blocking Screenshots to Protect Your Messages from AI Agents
Full Article
Original Source
Read the full article at Cnet →KhanList aggregates and links to publicly available news content. We do not host full articles from third-party sources. Always verify important information with original sources.