Signal's Meredith Whittaker warns that AI chatbots are surveillance tools, not companions

Signal President Meredith Whittaker has pushed back on the friendly framing of AI chatbots, calling them out as neither conscious nor trustworthy confidants.

In an interview with Bloomberg, Whittaker emphasized a clear distinction: "These are not your friends. These are not conscious beings. These are not sentient interlocutors." While she acknowledges using AI tools sparingly for formatting documents, she avoids offloading her own thinking to them, noting that relying on systems trained to average existing ideas risks foreclosing genuine intellectual work.

The privacy cost of convenience#

Whittaker took aim at recent industry predictions about AI handling everyday decisions. Microsoft AI CEO Mustafa Suleyman has suggested users could let Copilot handle holiday shopping by monitoring family conversations — a vision Whittaker sees as fundamentally misguided. Such a system would require granting an AI unprecedented access: credit cards, browser history, messaging apps, home addresses, and calendars.

"What you've just described is a system with very pervasive access across multiple applications and services," Whittaker said. "In the context of Signal, it would constitute a kind of a backdoor."

Her remarks underscore a recurring tension in AI adoption: the convenience of automation versus the surveillance infrastructure required to deliver it. As chatbots become increasingly woven into everyday services, her skepticism offers a counterpoint to the tech industry's enthusiasm about AI integration.

Read the full interview: TechCrunch — Signal's Meredith Whittaker wants you to remember that AI chatbots are not your friends