Meredith Whittaker: AI Chatbots Threaten Privacy

AI Chatbots Are Not Friends, According to Meredith Whittaker

Meredith Whittaker, president of Signal, recently expressed her concerns regarding the use of AI chatbots such as ChatGPT and Claude. In an interview with Bloomberg, she stated that these technologies are neither friends nor conscious beings, and should not be regarded as sentient interlocutors.

Whittaker shared that she occasionally uses AI tools for simple tasks, like formatting documents, but she avoids asking them important questions. She emphasizes the importance of preserving the integrity of her thinking and writing, without being influenced by responses generated by systems that merely reproduce existing data.

The Risks of Automating Personal Tasks

The president of Signal also reacted to a prediction by Mustafa Suleyman, CEO of Microsoft AI, that Microsoft Copilot could handle users' Christmas shopping. Whittaker criticized this scenario, pointing out that it would involve giving Copilot extensive access to personal information, such as credit cards, browsers, and even the ability to send personal messages. She mentioned that Copilot could spy on the family group chat to determine who wants what, and warned against such a system that, in the context of Signal, could be seen as a backdoor. She specifically highlighted the risks of access to her home address and calendar, thereby compromising the security and privacy of users across multiple applications and services.