ELON, September 13 — Half of all Americans now use large language models such as ChatGPT, Gemini, Copilot and Claude, according to a recent survey from Elon University’s Imagining the Digital Future Center. The finding, reported by HuffPost Life in a piece by Kelsey Borresen, lands as chatbots move out of the tech pages and into everyday routines — drafting messages, answering questions, sorting through problems that used to be handled by a search engine or a friend.
Lee Rainie, director of Elon’s Imagining the Digital Future Center, put the scale of the shift plainly in a university news release. “By any measure, the adoption and use of LLMs [large language models] is astounding,” Rainie said.
He added that he was “especially struck by the ways these tools are being woven into people’s social lives.” That last point is the one worth sitting with. A chatbot is not a search box. People type into it the way they talk to a confidant — and that is exactly the problem privacy and security specialists keep raising.
The HuffPost Life report, part of the outlet’s broader effort to sift health misinformation and bring readers facts that matter to their lives, frames the question directly: what is actually safe to hand over to these tools?
The categories experts say to keep out
The guidance centers on categories of personal information that users should avoid entering into ChatGPT and other AI chatbots. The report also addresses what users can do if they have already shared sensitive data — a recognition that many people have been typing into these tools for months without a second thought. The source material does not enumerate the specific items on that list, and this article will not guess at them.
What is confirmed is the shape of the advice: certain kinds of personal information should stay out of the chat window, and there are steps to take after the fact if they did not. The stakes are not hypothetical.
Rainie’s observation about social lives cuts both ways — the more naturally these tools fit into how people talk, confide and plan, the less likely anyone is to pause before hitting send.
What to watch
The Elon survey gives a clear baseline. Half of Americans are in. What remains unsettled is how much they understand about where their words go once they type them — and whether guidance like this reaches people before, not after, they have already shared something they wish they hadn’t.
