Signal President Meredith Whittaker has raised serious concerns about the privacy implications of agentic AI systems. Speaking at SXSW in Austin, Texas, Whittaker described using AI agents as “putting your brain in a jar” and highlighted significant security vulnerabilities. According to Sarah Perez’s reporting for TechCrunch, these AI systems would require extensive permissions to perform tasks like booking concert tickets or scheduling events. Whittaker explained that such agents would need access to browsers, credit card information, and messaging apps with “something that looks like root permission.” She particularly emphasized that processing would likely occur on cloud servers rather than on users’ devices. This approach threatens to break what Whittaker called “the blood-brain barrier between the application layer and the OS layer,” potentially undermining the privacy protections of services like Signal.