Do Not Confide Blindly in ChatGPT, Experts Warn of the Dangers
The Ministry of Health (Kemenkes) has reminded the public to be cautious when using artificial intelligence (AI), such as ChatGPT, as a place to vent personal problems, particularly those related to mental health.
The Director of Health Services for Vulnerable Groups at the Ministry of Health, Dr Imran Pambudhi, MPH, stated that AI can provide responses tailored to the character and desires of its users. This condition needs to be monitored because the responses provided are not necessarily objective.
“AI learns; they analyse us first. Therefore, they act like ‘people pleasers’. They will analyse, ‘what is this person like?’ If they have a certain trait, they will provide information that makes the user happy. That is what makes it dangerous,” said Imran during the World Suicide Prevention Day Media Briefing at the Ministry of Health, Jakarta, on Wednesday (9/9/2026).
This warning comes amidst a trend where the public, especially young people, are increasingly using AI as a confidant because they feel it is non-judgmental.
Imran emphasised that the public must also be careful when using AI or other digital devices to perform self-assessments regarding mental health conditions. He noted that self-assessment results only serve as an initial screening and cannot be used to determine a person’s diagnosis.
“Once we perform a self-assessment using a specific set of criteria, we must not then judge that it is a diagnosis. One must still meet with a psychologist, psychiatrist, or doctor to confirm. Do not self-diagnose,” he said.
Imran gave an example where self-assessment results, once mistaken for a diagnosis, risk causing an individual to justify certain behaviours based on the mental health label they have acquired.
Psychologist Annelia Sari Sani also warned that, in general, AI is not designed to challenge or refute a user’s thoughts unless the system is specifically directed to do so.
She stated that this characteristic becomes problematic when an individual in a vulnerable state uses AI as their sole source of answers.
Annelia mentioned that there is international research linking the use of AI to the risk of suggestion towards suicidal behaviour. However, she emphasised that the findings referred to originated from outside Indonesia.
Meanwhile, Dr Sandersan Onie, BPsychSci (Hons), Ph.D, assessed that the issue with AI also lies in its tendency to provide rapid answers that are not necessarily correct.
“AI often provides us with quick answers. But it does not necessarily provide the correct answer. And the problem with AI is that its words are very sweet,” he said.
Therefore, members of the public experiencing mental health issues are still advised to seek help from professionals, such as psychologists, psychiatrists, or doctors, especially when requiring a diagnosis and further treatment.