Users delete personal AI agents over privacy concerns and security issues

Early adopters of personal AI agents like Instinct and Meta's Muse are increasingly deleting or restricting their use due to privacy concerns and security incidents. Four users, including Guto Martino of Hermes Agents Brasil, cited uncertainty about data handling and incidents such as unauthorized access to Gmail login codes and suspicious activity from Iran. Meta defended Muse's privacy measures, emphasizing user control over data and connections, while Instinct did not respond to inquiries. Scott Persinger, CTO of BizTrip, deleted Instinct over concerns about email access, preferring safer alternatives. Rami Elghandour, CEO of Arcellx, deleted Muse after reports of unauthorized text message access, opting for self-built agents instead. Mahesh Vellanki, founder of YieldClub, deleted Instinct after a suspicious two-factor authentication request from an Iranian IP address. Despite these concerns, some users like Rishi Bhargava continue to experiment with agents but with limited access. Companies like Instinct and Meta emphasize data isolation and user control, though users demand greater transparency and explicit authorization for agent actions.