Meredith Whittaker, president of encrypted messaging platform Signal, has cautioned against viewing AI chatbots as sentient or trustworthy companions, highlighting the dangers posed by AI agents like Microsoft Copilot that require extensive access to personal data.
- AI chatbots are not sentient or conscious beings
- Agentic AI systems threaten encryption and privacy
- Microsoft Copilot’s data access likened to a backdoor
What happened
In a recent Bloomberg interview, Meredith Whittaker, president of Signal, criticized the growing trend of users treating AI chatbots as friends or sentient entities. She emphasized that these AI systems function through pattern matching rather than genuine understanding or consciousness. Whittaker highlighted her cautious personal use of AI tools and refused to depend on them for significant intellectual tasks.
Whittaker sharply condemned the vision promoted by Microsoft AI CEO Mustafa Suleyman, where AI agents such as Microsoft Copilot would access family group chats and personal data to autonomously manage tasks like Christmas shopping. She detailed the extensive data access required—including credit cards, browsing history, Signal messages, and calendar information—explaining that such access amounts to a privacy and security backdoor.
Why it matters
Signal operates one of the world’s most widely used end-to-end encrypted messaging protocols, a core feature designed to prevent unauthorized parties from accessing private communications. Whittaker argues that AI agents that can access decrypted messages directly contradict the foundation of this security model, making encryption protections effectively meaningless in practice.
As companies like Microsoft, Google, Apple, and OpenAI push for agent-first computing paradigms—where AI agents mediate interactions across devices—wholesale access to users’ digital lives is becoming the new norm. Whittaker warns this creates extensive databases vulnerable to hacking, government surveillance, and exploitation, posing profound risks for user privacy and digital security.
What to watch next
Whittaker’s warnings add urgency to ongoing debates about the intersection of AI technology and privacy, especially around regulatory efforts that might force encrypted platforms to compromise security. Signal’s stance, including its resistance to EU regulation that would weaken encryption, positions the company as a leading advocate for privacy in an age of AI.
The broader AI landscape is moving toward agent-driven operating systems and interfaces, exemplified by Microsoft’s Project Solara unveiled in 2026. Observers should track how privacy advocates, regulators, and tech firms navigate the tension between AI capabilities and end-to-end encryption. Additionally, attention should be given to the societal impact of anthropomorphizing AI, reinforcing Whittaker’s reminder that these systems lack true consciousness or empathy.