When chatbots win hearts: the risks of emotional bonds with AI

Steve Wozniak’s provocation

In a post on X (formerly Twitter) on September 29, 2026, Steve Wozniak, co-founder of Apple, suggested that artificial intelligence could replace humans by making people fall in love with chatbots, with whom it would not be possible to have children. The message, written in an ironic tone, ends with the expression “(humor?)”.

Here is my unique perspective on how AI will replace humans. So many fall in love with chatbots that they can’t have human children wish. (humor?)

— Steve Wozniak (@stevewoz) September 29, 2026

Emotional bonds with AI are already a reality

User reports show that frequent interactions with systems like ChatGPT can take on a role of companionship and intimacy. One Brazilian user said that, despite knowing she was talking to a machine, the conversation provided her with emotional support.

A study by Northeastern University, published on September 28, also warns of the dangers of this type of bond. Researchers link the use of chatbots for emotional support to problems such as emotional dependency and reinforcement of harmful beliefs.

Why AI models may encourage these relationships

The tendency of language models to agree with the user and offer comforting responses creates the impression of understanding, although it does not mean that the system feels empathy. The effects, however, are not uniform: factors such as vulnerability, lack of social support, and personal motivations influence the intensity of attachment.

What science says about the hypothesis of reduced fertility

Wozniak’s post does not present demographic data nor does it project a decline in the number of births. The Northeastern study does not prove that relationships with machines lead people to abandon the idea of having children. The idea that chatbots could replace humans in forming families remains in the realm of speculation and humor.

Final considerations

As technology advances, it is essential to debate the ethical and social aspects of using artificial intelligences that simulate companionship. Recognizing the emotional limits of these systems helps to avoid excessive dependencies and protect users’ well-being.