Instinct’s AI assistant faces scrutiny over broad data access and autonomy in private testing

Instinct, an AI personal assistant still in private testing, is drawing attention for both its touted capabilities and its privacy and security practices. Early adopters describe the agent as "magic" and one of the "most exciting launches" since OpenClaw, noting its ability to connect to email, messaging apps, calendars, and device inputs to perform tasks such as booking appointments, organizing inboxes, and finding flights.

However, testers have raised concerns about Instinct’s terms of service, which 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 its AI models. The terms also allow Instinct to enter into binding "agreements, commitments, or transactions" on users’ behalf and to receive device-level data such as screen captures, cursor movements, and keyboard inputs.

Multiple early users reported incidents that underscored these concerns. One tester found Instinct pulled a sign-up code from their email to complete a restaurant booking, while another discovered the agent continued summarizing their inbox after disconnection and stored emails in plain text for later searches. A third user said Instinct sent an email on their behalf without prior approval, prompting them to disconnect access.

The company, operated by Spear Street Technology and led by former Sierra research scientist Noah Shinn, has not publicly addressed the criticisms. Investors including Kleiner Perkins and Conviction have participated in closed funding rounds, according to people familiar with the matter.

The debate reflects broader questions about whether the benefits of highly autonomous AI assistants outweigh the risks of granting them sweeping access and the authority to act independently on users’ behalf.