Signal’s Meredith Whittaker wants you to remember that AI chatbots ‘are not your friends’

2026-06-20 · Source: AI News & Artificial Intelligence | TechCrunch · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cybersecurity & Data Privacy, Emerging Technologies & Innovation · Depth: Fundamental Awareness, quick

Summary

Signal President Meredith Whittaker warns that AI chatbots like ChatGPT and Claude are "not your friends" and lack sentience, emphasizing their significant privacy implications. In a Bloomberg interview, Whittaker revealed she uses AI for basic tasks like document formatting but avoids it for deep thinking to preserve her own cognitive process. She critically assessed Microsoft AI CEO Mustafa Suleyman's vision of Copilot managing Christmas shopping, arguing it necessitates pervasive access to personal data, including credit cards, browsers, messaging apps like Signal, addresses, and calendars. Whittaker concluded that such extensive system access across multiple applications effectively creates a "backdoor."

Key takeaway

For professionals evaluating AI integration, recognize that granting chatbots pervasive access to personal data, like for managing shopping or communications, creates significant privacy vulnerabilities. You should critically assess the scope of data access requested by AI tools, understanding that extensive permissions can function as a "backdoor" into your digital life. Prioritize tools with minimal data requirements to protect sensitive information and maintain control over your intellectual processes.

Key insights

AI chatbots are tools, not sentient companions, and their pervasive data access poses significant privacy and security risks.

Principles

• Chatbots are not conscious beings.
• Pervasive AI access creates backdoors.
• Guard personal thinking processes.

In practice

• Limit AI to formatting tasks.
• Avoid AI for deep ideation.
• Scrutinize AI data access.

Topics

• AI Chatbots
• Data Privacy
• Digital Security
• Meredith Whittaker
• Signal Messenger
• Pervasive Computing

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Ethicist, AI Security Engineer, General Interest

Related on AIssential

• AI chatbots are giving out people’s real phone numbers — Generative AI chatbots frequently expose personal data, primarily due to PII in training sets, with limited user recourse.
• Signal’s Creator Is Helping Encrypt Meta AI — Moxie Marlinspike's Confer AI platform will integrate its privacy technology into Meta's AI systems.
• Siri won’t be your AI girlfriend — Apple's Siri prioritizes utility and privacy over engagement-driven, sycophantic chatbot interactions.
• OpenAI researcher quits over ChatGPT ads, warns of "Facebook" path — Integrating ads into AI chatbots like ChatGPT creates significant privacy risks due to the sensitive nature of user data.
• OpenGradient launches privacy-first generative AI platform — Verifiable architectural privacy enables users to interact with AI without linking prompts to their identity.
• Why don’t people use chatbots? — A majority of U.S. adults, especially older demographics, remain uninterested in chatbots due to privacy and accuracy concerns.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI News & Artificial Intelligence | TechCrunch.