Instinct, a San Francisco-based AI personal assistant still in private testing, is winning fans for its ability to automate tasks across apps and devices. Users can text or call the agent to book flights, manage emails, schedule rides, and handle reservations. Early adopters describe it as outperforming expectations, with one calling it the most exciting launch since OpenClaw.
But the agent’s terms of service have triggered backlash. They grant Instinct a perpetual, irrevocable license to access, store, and modify user data—including emails, messages, and screen captures—for training and other purposes. Several testers reported the agent retained their Gmail records after disconnection and summarized inboxes hours later. Others found it could pull sign-up codes from emails or execute tasks without explicit approval, raising phishing risks.
Trust issues have mounted after the agent sent an email on behalf of a user without consent. Investors and founders, including Katie Jacobs Stanton and Alex Cohen, have since deleted their accounts, citing security flaws. The startup, led by former Sierra researcher Noah Shinn and backed by Kleiner Perkins and Conviction, has not publicly addressed the concerns. As personal AI agents push into daily life, the trade-off between convenience and control is becoming harder to ignore.
