Signal President Meredith Whittaker has warned users that AI chatbots like ChatGPT and Claude are not conscious friends, speaking in a recent Bloomberg interview about the severe privacy risks of integrating artificial intelligence into daily life.
The Danger of Outsourcing Human Thought
During a broader interview with Bloomberg focusing on policy, privacy, and her work at Signal, Whittaker revealed she only uses AI tools occasionally to format documents. She strictly avoids asking chatbots questions to protect her intellectual independence.
The tech executive emphasised that she refuses to let her own thinking and writing processes be eclipsed or foreclosed by systems designed to merely average out existing online information, stating she remains highly serious about her own cognitive process.
Microsoft Copilot and the ‘Backdoor’ Threat
Whittaker also addressed a prediction by Microsoft AI CEO Mustafa Suleyman, who suggested users could soon delegate all their Christmas shopping to Microsoft Copilot. She highlighted the invasive reality of such convenience, noting that allowing an AI assistant to handle shopping means granting it sweeping access to personal credit cards, web browsers, home addresses, calendars, and private communication platforms.
A Perilous Level of Access
According to Whittaker, this level of integration creates a highly pervasive system that spans multiple services and applications. In the context of secure messaging platforms, she warned that granting an AI such deep access would ultimately constitute a dangerous backdoor, compromising the fundamental privacy that apps like Signal are built to protect.