A chatbot may sound caring, give advice, and seem easy to trust. The AP Stylebook now says journalists should remember that AI systems “do not think, feel, want or understand.” That advice lands differently when Gen Z is described as forming emotional bonds with bots.

The guidance is part of a wider discussion about how newsrooms should write about AI. The AP Stylebook says reporters should avoid giving AI human traits. It asks them to explain what a system does, how well it works, who built it, and who may be affected.

That sounds simple. It is not.

People use human words because the systems often produce human-sounding replies. A bot can answer with warmth. It can respond at any hour. It can offer advice without showing impatience. The words may feel personal even when the system has no feelings behind them.

This is where the debate becomes a local concern, not only a newsroom rule. Readers meet AI in schools, offices, homes, and phones. The language used by journalists helps shape what people expect from these tools. If a report says a bot “understands” a user, what does a reader hear? If it says the bot “wants” to help, who carries the blame when that help is wrong?

A LinkedIn discussion put the concern in generational terms. Libby Rodney argued that Gen Z is “the most AI-native generation alive” and is forming emotional bonds with bots for comfort and advice.

The claim is striking. It is also a claim, not a complete picture. What does “emotional bond” mean here? Does it mean a habit, a sense of comfort, trust, or something closer to friendship? How was that feeling measured? And why is this framing appearing now, when companies are placing more AI tools in daily life?

Those questions matter because the words can move attention away from the system and toward the character it appears to play. A bot may seem kind, wise, or loyal. But the AP guidance points writers back to the people who designed and deployed it.

That distinction is not small. If a system gives unsafe advice, a phrase such as “the bot made a bad choice” can make the event sound like the result of a personal act. The real questions are harder. Who set the rules? Who tested the system? Who approved its use? Who is responsible for the result?

Allen Holub and Stephen Klein have stressed human accountability in the discussion. They have also argued that framing models as actors can affect safeguards for their behavior. That concern deserves room beside the claims about comfort and companionship.

But there is another risk. If writers avoid every human word, will they also fail to describe what users actually experience? A person can know that a bot has no feelings and still feel comforted by its reply. Both facts can exist at once. The first describes the system. The second describes the person.

This is why the debate should not be treated as a fight between cold facts and warm human needs. The language can be careful without pretending that users feel nothing. A report can say that a bot generated supportive words. It can also say that a user experienced those words as support, if that experience is documented.

The difference is who owns the feeling.

The bot does not need to be called lonely, caring, afraid, or proud. Those words belong to people. Giving them to software may make a story easier to read, but it can also make a product seem safer or wiser than the evidence allows.

Why does this story exist? Is it about a style rule, a generational shift, a business opportunity, or a safety problem? Who benefits when a tool is described as a companion rather than as a system that produces replies?

No single wording settles the issue. The AP Stylebook offers a clear rule for journalists, but the public question remains open: how should we describe a technology that can imitate closeness without possessing a human inner life?

For readers, the safest habit may be an old one. Pay attention to what the system does. Ask who built it. Ask who gains from its use. And when a bot sounds like it cares, remember that the feeling may be real for the person reading the words, even if it does not belong to the machine.

More on: Local News