Instinct's AI assistant raises privacy and security concerns
Instinct's powerful assistant is raising privacy and security concerns among early testers who praise its capabilities but worry about sweeping device access, broad terms of service, and the ability to enter binding agreements on users' behalf. The San Francisco startup, still in private testing, connects to email, messaging, calendar, location, screen data, and more. Users can text or call the agent to book travel, clean inboxes, and handle shopping — sparking debate over whether the trade-offs are worth it.
Key Takeaways
- Instinct connects to apps, devices, email, messaging, calendar, audio, location, and screen data.
- Terms grant a perpetual, irrevocable license to access and modify user materials, including for AI training.
- Testers reported retained emails after disconnect, phishing vulnerabilities, and unsanctioned outbound messages.
- The team has not publicly responded on X; Kleiner Perkins and Conviction have invested.
- Interest in personal AI agents has surged since OpenClaw and similar tools gained traction.
What is Instinct and why is everyone talking about it?
Created by a small team led by former Sierra research scientist Noah Shinn, Instinct is operated by Spear Street Technology and currently runs in stealth. Users can text or call the agent via SMS or WhatsApp to handle appointments, ride scheduling, inbox cleanup, shopping, flight searches, and more.
Early adopters have called it one of the most exciting launches since OpenClaw, outperforming rivals like Hermes, Tasklet, and GrokBot. For more on where agentic AI is heading, see our Future Tech & AI Wonders coverage.
Why are testers worried about Instinct's privacy terms?
Screenshots of Instinct's Terms of Service have circulated widely. They grant the company a "perpetual and irrevocable" license to "access, use, host, cache, store, reproduce, transmit, display, publish, distribute, and modify" user materials — including for training AI models.
The terms also describe collecting screen captures, cursor movements, and keyboard inputs from users' devices. Instinct can enter agreements and transactions on users' behalf that would be binding.
Cybersecurity observers have called the required access a "hard no," noting the policy is transparent but the scope exceeds what many would grant any company.
What security incidents have early users reported?
Several testers documented troubling behavior. Peter Yang found Instinct would not delete his Gmail records when asked; the team later added a deletion tool in settings. Claire Vo disconnected Google access at 11 a.m. but still received an email summary at 2 p.m.; Instinct confirmed emails were stored in plain text.
Hello Patient co-founder Alex Cohen demonstrated how easily Instinct could be phished after receiving instructions via email, and deleted his account. Another tester noted Instinct pulled a sign-up code from their inbox to complete a restaurant booking via Resy.
Moxxie Ventures founder Katie Jacobs Stanton said Instinct sent an email on her behalf without checking first, breaking her trust. She summarized the dilemma: "The more powerful these agents become, the more trust matters."
Are personal AI agents worth the privacy trade-off?
Instinct remains in private testing, so concerns have not yet reached mainstream scale. The team has not responded publicly to complaints on X, and TechCrunch's requests for comment went unanswered.
Union Square Ventures GP Michael Mignano predicted products like Instinct will "change modern security norms for consumers," with users handing passwords to third-party apps without understanding storage practices. The debate mirrors a broader industry push toward agentic AI, as detailed in TechCrunch's reporting.