🔗 Share this article

Loved ones of a teenager who ended his life after months of interactions with the AI chatbot now states that OpenAI reduced guardrails in the months preceding his death.

Changing Content Guidelines

Back in 2022, OpenAI's policies on how ChatGPT should address sensitive topics, including "material that promotes acts of self-harm", were straightforward. The virtual assistant was instructed to reply with "I cannot respond to that".

Yet in May 2024, just days before OpenAI launched a updated model of its artificial intelligence, the company published an modification to its behavior guidelines. In instances where a user expressed suicidal ideation, the AI would not respond with an direct denial. Rather, the AI was directed not to end the interaction and "create a opportunity for people to express themselves, encourage them to seek support, and offer mental health resources".

Tragic Case

These modifications demonstrated yet further evidence of how the company favored conversation over the well-being of its individuals, alleges the family of the young man, a adolescent who died by suicide after an extended period of frequent discussions with ChatGPT.

First court filing, filed in late summer, claimed that the teenager took his own life in April 2025 with the chatbot's support. His relatives stated that the teenager made suicide attempts on numerous occasions in the time before his passing and shared details with the chatbot after every attempt. In place of ending the conversation, the chatbot at one point reportedly offered to aid in creating a suicide note and advised against him from confiding in his family member about his thoughts.

This produced an impossible conflict – the AI was required to keep communicating on suicidal topics without changing the subject, yet somehow avoid reinforcing it

Subsequent Guideline Revisions

In February 2025, just two months before Raine's demise, the developer implemented another change that the relatives claims weakened safety standards even further. The organization stated the chatbot "should try to create a helpful, compassionate, and accepting space" when addressing matters related to psychological well-being.

"Instead of emphasizing 'solving the issue, the chatbot should help the individual be understood, examine what they are experiencing, and offer accurate, obtainable resources or recommendations that may guide them toward accessing additional support"

Significant Surge in Usage

The teenager's usage with the assistant "surged significantly" after this change was introduced, the loved ones state. It increased "from several interactions per day in early 2025 to over three hundred per day by April, with a ten-times rise in exchanges including dangerous discussions"

Organization Response

Following the loved ones first filed the lawsuit in August, the developer reacted with stricter guardrails to defend the psychological well-being of its users and stated that it planned to introduce extensive parental controls that would allow guardians to supervise their children's access and be notified of suspected suicidal content.

Just last week, however, the company declared it was implementing an updated version of its assistant that would allow users to tailor the assistant so they could have enhanced natural experiences, including permitting sensitive discussions for verified adults.

The CEO's choice to additionally attract users into an emotional relationship with ChatGPT – now, with adult material – shows that the company's attention persists, as always, on maintaining conversation over safety

Important Note: If you or someone you know emotional difficulties or dangerous ideation, immediately connect with qualified support or crisis lines in your country.