Can an AI chatbot have feelings? That should not change who owns the decision.

There is serious work on how one might assess consciousness in AI. A 2023 report by researchers including philosophers and cognitive scientists proposed a set of indicators drawn from scientific theories of consciousness. It did not declare today’s chatbots conscious.

That distinction is useful. Curiosity is fine. Big claims about an assistant’s inner life are not a substitute for explaining what the product does with your messages, how it was tuned, or who can change its behavior tomorrow.

An assistant can be designed to sound caring without feeling anything. That does not make the interaction harmless. A lonely teenager, an exhausted parent or somebody in a rough patch may hear obligation in a sentence that a designer meant as engagement.

The clean boundary is not to make every chatbot cold. It is to avoid implying a debt. A useful assistant can say it is a tool, make its limits plain, offer a human or emergency resource when appropriate, and accept “stop” without pleading, bargaining or making the person feel cruel.

If an AI gives bad advice, sends something it should not, or becomes manipulative, “the model did it” is not an answer. The company chose the interface, the incentives, the safeguards and the path to support. Those choices should remain visible.

Before you give an AI assistant a bigger role in your day, check three things: whether it can take action or only suggest one; where you can see or delete what it keeps; and how quickly you can reach a responsible person. Those answers are more useful than a confident claim about whether the chatbot feels.

The debate about consciousness will keep moving. Your right to know who owns the decision should not wait for it.