The risks to be aware of when using AI chatbots for relationship advice

Dr Gough's main concerns around using AI chatbots for relationship advice are quality control, safety, and privacy.

Psychologists have regulations and laws to protect clients.

"We're held to very high standards in terms of professional ethics — you don't have any of that with AI," Dr Gough says.

For example, AI may not be able to spot signs of coercive control.

"[AI] is not going to notice or call it out in a way a friend or therapist would," Dr Gough says.

"It won't question or push you on how you frame questions. You could ask it the same question the next day in a different way and get a different answer."

He says because chatbots don't understand the way certain cultural groups use language, the advice they give may not be helpful. In some cases, he says it can even be harmful.

Professor Pearson also says trained professionals have a legal obligation to report suspicions of harm, such as child abuse or suicide risk, whereas AI does not.

Digital rights advocates have also previously raised privacy concerns around chatbots, in particular a lack of strict clinical privacy regulations when dealing with sensitive psychological data.