Meredith Whittaker, president of the encrypted messaging service Signal, has warned against treating AI chatbots as social companions. Instead, she said they should be viewed as software systems that generate responses, not as entities capable of friendship or awareness.
Her remarks came during a Bloomberg interview, as reported by TechCrunch on June 20, 2026. The discussion focused on privacy and the growing role of AI assistants such as ChatGPT and Claude.
“These are not your friends. These are not conscious beings. These are not sentient interlocutors,” Whittaker said.
She also noted that she uses AI tools in limited ways. For instance, she relies on them to format documents. However, she avoids deeper interaction. “I don’t ask them questions,” she said. She added that she prefers to work through ideas without machine-generated responses shaping her thinking.
As a result, she raised concerns that frequent reliance on chatbots could shift how people develop ideas and structure thought.
Privacy concerns around expanding AI systems
Whittaker also addressed comments from Microsoft AI CEO Mustafa Suleyman about future AI agents that could handle tasks like holiday shopping.
In that scenario, she said, such systems would require broad access to personal data across multiple services. Because of this, she warned that deeper integration could create serious privacy risks.
She referenced systems like Microsoft Copilot as examples of tools that could operate across platforms and user accounts.
“What you’ve just described is a system with very pervasive access across multiple applications and services,” she said. She added that, in Signal’s case, this level of access would resemble a “backdoor.”
Her concern focused on the permissions required for AI agents to reach messaging apps, email, calendars, and financial data.
Balancing convenience and control
The comments highlight an ongoing debate in the tech industry. On one side, companies are building AI systems that automate tasks across apps and services. On the other, privacy advocates warn about the risks of granting such broad access.
In addition, Whittaker said these systems may improve convenience but also expand the amount of sensitive data processed by a single platform.
Therefore, she urged caution in how much control users give up in exchange for automation.
The debate reflects growing tension between AI-driven convenience and data protection as assistants become more embedded in daily digital life.
Tags: Signal, Meredith Whittaker, AI chatbots, Privacy, Cybersecurity, Microsoft Copilot, ChatGPT, Claude AI, Data security, Agentic AI
