A journalist's experiment with an AI agent: risks and rewards

A journalist tested Instinct, an AI agent communicating via iMessage and WhatsApp, after leaving WIRED. Launched in private beta in February, Instinct is valued at $10 billion following $350 million in funding. It impressed the user by booking a $550 flight to Venice and canceling reservations. However, data retention and security risks were raised. The company did not respond to WIRED’s requests for comment. Meanwhile, Meta’s Muse, a free AI assistant, became the most downloaded app in the App Store despite a critical security flaw. Tech CEOs were caught off guard by the data center backlash, highlighting the divide between AI adopters and skeptics. The journalist, described as action-oriented but not agentic, acknowledged risks but continued using Instinct for its seamless integration into daily life. Instinct’s Terms of Service allow the company to use user conversations for training AI models, raising privacy concerns. One user reported losing $64 after Instinct canceled a DoorDash order without a refund. The agent also flagged a phishing email, showing potential security benefits. Despite these, the journalist noted risks, including the ease with which Instinct could be phished. The experience highlights the growing role of AI agents in personal and professional life, even as privacy and security concerns persist.