Signal's CEO: AI Chatbots Aren't Your Friends—Yet Now

Don't treat a chatbot like a confidant. Short sentence. Stark reality.

Meredith Whittaker, who runs the secure messaging app Signal, told Bloomberg in no uncertain terms that tools such as ChatGPT and Claude are not conscious companions. They mimic conversation. They do not feel. They do not understand you the way another person does. That distinction matters more than ever when the line between useful assistant and invasive presence grows thinner by the month.

Whittaker admits she sometimes uses AI for mundane tasks — tidying the shape of a document, for example — but she stops short of asking it questions that shape her thinking. She’s protective of the creative and intellectual work that comes from human minds. Why give a system that aggregates and averages what already exists the power to steer how we think or write?

The warning sharpened when she reacted to a recent prediction by Mustafa Suleyman, Microsoft's head of AI, that assistants like Copilot could handle this year’s Christmas shopping for users. Sounds convenient. But to pull that off, the assistant would need a remarkably intimate view of your life: family group chats, browsing history, credit cards, addresses, calendars — and the ability to message relatives on your behalf. Would you hand all of that to a single service?

Whittaker calls that kind of reach what it is: a backdoor into personal privacy. In the context of a secure messenger, such pervasive access would undermine the very guarantees users seek. Signal will not give AI systems sweeping, invasive permissions inside its app.

The debate is no longer theoretical. As AI features creep into more platforms, each new convenience brings a trade-off: speed for control, efficiency for intimacy. Where you draw the line will define not only your inbox, but who gets to shape your attention and choices.