Security Risk: Signal CEO Meredith Whittaker Warns Against AI Chatbots
Signal CEO Meredith Whittaker has criticized the excessive use of AI chatbots and agents, warning they could pose significant security risks by undermining end-to-end encryption.
- Home
- News
- Software & Development
Don't miss any news!
Signal CEO Meredith Whittaker has spoken out against the excessive use of AI chatbots and AI agents. These could pose a massive security risk, for instance, by undermining end-to-end encryption.
For years, Meredith Whittaker, president of the Signal Foundation, has warned about the AI hype and potential dangers to data security. Therefore, the encrypted messenger Signal consciously refrains from integrating chatbots or AI features.
AI Chatbots as a Security Risk
In a detailed interview with Bloomberg, Whittaker elaborated on why she specifically considers AI chatbots and AI agents to be a significant security risk.
Whittaker herself uses chatbots like ChatGPT, but only occasionally to format a document. She avoids conversations with AI chatbots, stating that she takes her thinking and writing very seriously and does not want the process of idea generation to be hindered or overshadowed by a "system that merely determines the average of existing data."
Whittaker: Chatbots Are Not Friends
For regular chatbot users, Whittaker has a clear warning: "They are not your friends. They are not conscious beings. They are not sentient conversational partners," the tech expert told Bloomberg. Furthermore, chat histories are also stored, logged, and searched for data, according to Whittaker.
In particular, a vision of the future from Microsoft's AI chief Mustafa Suleyman made her reflect. Suleyman had stated in late December 2025 that users could complete their Christmas shopping entirely via Copilot. However, this would mean that Copilot would need to eavesdrop on family chats to learn the wishes of family members.
Extensive Access Required
Moreover, the AI tool would require access to users' credit cards, browsers, home addresses, and calendars. This system would thus have extensive access to numerous applications and services, Whittaker noted. Additionally, Copilot would represent a kind of backdoor for Signal.
An AI agent integrated into an operating system with the appropriate rights could intercept content in plaintext before or after encryption, as reported by heise.de.
Tech Giants Control Operating Systems
And: The central operating systems are controlled by only three companies, namely Microsoft (Windows), Google (Android), and Apple (iOS). These companies, according to Whittaker's concerns, could "make decisions by decree that fundamentally harm collective cybersecurity."
