As AI tools become more conversational and deeply integrated into daily life, privacy experts warn users against treating them as trusted companions.
The rapid rise of AI chatbots like ChatGPT and Claude has changed how people interact with technology. From answering questions to offering emotional support, these systems are increasingly being used as personal companions, advisors, and productivity tools. But according to Meredith Whittaker, that growing dependence comes with serious risks.
Speaking recently about the evolving role of AI, Whittaker cautioned users against humanizing these systems, reminding them that AI chatbots are “not your friends” and should not be mistaken for conscious or trustworthy companions. Her remarks reflect a broader concern within the tech and privacy community: as AI becomes more conversational, users may begin to overlook the data collection and surveillance systems behind it.
The Privacy Trade-Off Behind Convenience
Whittaker, who leads the encrypted messaging platform Signal, has long been vocal about digital privacy. Her concerns focus on how AI assistants often require access to personal data, emails, calendars, messages, and even financial information to function more effectively.
While this integration makes AI more useful, it also creates deeper data dependencies. She has particularly criticized the growing trend of “agentic AI,” where systems can act on behalf of users across multiple apps. According to Whittaker, this level of access could create privacy vulnerabilities that many users may not fully understand.
A Bigger Debate Around Trust and Technology
Whittaker’s warning comes at a time when AI companionship is becoming a mainstream phenomenon. Studies show users are increasingly forming emotional attachments to chatbots, blurring the line between utility and relationship. But experts argue that these systems are fundamentally designed to predict responses, not care, empathize, or protect users.
As AI becomes more embedded in personal and professional life, the larger question remains: how much trust should users place in systems built by corporations whose core model often revolves around data?
Whittaker’s message is clear: AI may be useful, but understanding its limits could be just as important as embracing its capabilities.