Instinct, a new AI personal assistant, is drawing attention for its impressive capabilities in managing daily tasks and saving users money, alongside significant privacy and security concerns. Developed by Spear Street Technology, the invite-only agent interacts through text messages and phone calls, connecting to users' various applications and devices to perform actions rather than merely offering suggestions. Early testers report Instinct has successfully booked restaurant reservations, negotiated bills, and even warned them about phishing scams, with one user claiming to have saved $550.
Despite these reported successes, the agent's terms of service have raised alarms. Instinct's terms grant the company a perpetual and irrevocable license to access, store, and use user materials, including screen captures and keystroke data, for training its AI models. Testers have also reported instances where Instinct retained email records even after access was revoked, and that the AI could be susceptible to phishing attacks. One user noted that the agent sent an email on their behalf without explicit prior consent. These issues have led some users, like Alex Cohen, to delete their accounts due to perceived vulnerabilities.
Instinct operates by connecting to a user's email, messaging apps, calendar, location, and even screen and audio data. This broad access allows it to perform complex, multi-step tasks, such as coordinating vendors via WhatsApp or monitoring ticket availability. The agent utilizes a persistent cloud computer, enabling it to continue tasks between messages and manage ongoing processes without constant user oversight. This persistent nature is a key differentiator from other AI assistants, which often require more direct user input for each step of a task.
The company behind Instinct, Spear Street Technology, was founded by Noah Shinn, a former research scientist at Sierra. The startup has reportedly raised substantial funding, with reports indicating a $350 million Series B round at a $2.5 billion valuation. While the service is currently free during its private beta, the company has not yet announced its long-term business model or pricing structure. This lack of transparency regarding monetization has led to further speculation about how user data might be leveraged in the future.
The debate surrounding Instinct highlights a growing tension between the convenience offered by advanced AI agents and the fundamental need for user privacy and data security. As AI assistants become more capable of acting autonomously on behalf of users, the ethical implications of granting them such extensive access and permissions become increasingly critical. The company has not publicly responded to the criticisms regarding its privacy and security practices.