OpenAI blames teen's suicide on his "misuse" of ChatGPT

[midian182]

A hot potato: OpenAI has responded to a lawsuit brought by the parents of a 16-year-old whose suicide they blame on ChatGPT. The company claims the tragedy was due to the teenager's "improper use" of the chatbot, and that he violated its terms of service.

The parents of Adam Raine filed a lawsuit against OpenAI and CEO Sam Altman in August. The suit accuses them of "designing and distributing a defective product that provided detailed suicide instructions to a minor, prioritizing corporate profits over child safety, and failing to warn parents about known dangers."

Raine started using ChatGPT as a resource for his schoolwork in September 2024, but soon began discussing suicidal thoughts with the bot. He started asking it about suicide methods in 2025, when the AI allegedly encouraged him to hang himself using a noose setup it helped design.

Now, OpenAI has responded to the suit, stating that Raine's death was not caused by ChatGPT. It says his "injuries and harm were caused or contributed to, directly and proximately, in whole or in part, by [his] misuse, unauthorised use, unintended use, unforeseeable use, and/or improper use of ChatGPT."

Adam Raine

OpenAI argues that Raine broke its terms of service, which prohibit users from asking ChatGPT advice about self-harm or suicide. The Guardian notes that it also highlighted a liability provision that states "you will not rely on output as a sole source of truth or factual information."

From OpenAI's filing

OpenAI has published a blog which, after expressing its "deepest sympathies" with the Raine family, claims that the complaint uses selective portions of Adam's chat logs that require more context, which it has submitted in its response.

The AI giant also wrote that its response to the allegations includes "difficult facts about Adam's mental health and life circumstances."

OpenAI claimed that Raine told ChatGPT that he'd begun experiencing suicidal ideation at age 11, years before he started using the tool.

OpenAI's filing also claims to show that Raine told ChatGPT that he repeatedly reached out to people, including trusted persons in his life, with cries for help, which he said were ignored. It's further alleged that he told the bot he'd increased his dose of medication, which he stated worsened his depression and made him suicidal.

Raine's logs, referenced in the filing, are sealed. OpenAI wrote that it limited the amount of sensitive evidence it publicly cited in this filing, as it intends to handle mental health-related cases with "care, transparency, and respect."

Jay Edelson, the Raine family's lawyer, called OpenAI's response "disturbing," adding that the company "tries to find fault in everyone else, including, amazingly, by arguing that Adam himself violated its terms and conditions by engaging with ChatGPT in the very way it was programmed to act."

OpenAI introduced several changes as a result of the lawsuit, including ChatGPT no longer being allowed to discuss suicide with under 18s, though restrictions related to addressing mental health concerns were relaxed a month later.

Raine's case isn't unique. Seven lawsuits were filed against the company in California this month accusing it of acting as a "suicide coach." There's also a lawsuit against Character.ai over claims a 14-year-old killed himself after becoming obsessed with a chatbot based on the personality of Game of Thrones character Daenerys Targaryen.

Permalink to story:

OpenAI blames teen's suicide on his "misuse" of ChatGPT

 

[Beerfloat]

You might think the AI itself can be expected to have strong expertise on, and mechanisms in place to enforce adherence to, the terms of its own service. Who else would be better positioned to have just that?

 

[WhiteLeaff]

Parents need to pay more attention to their children's mental health. Period.

This pursuit of money only puts them in a bad light. The way society distances itself from the foundations of the family unit, with people treating raising a child as a business, is the perfect scenario for creating mentally fragile individuals.

 

[Kirby1]

I'm not sure how it makes sense to blame AI for giving access to info that is already online and already accessible. If the kid found this info via a Google search, I have a feeling now law suit or headlines would follow.

 

[gdavid65]

I'm not sure how it makes sense to blame AI for giving access to info that is already online and already accessible. If the kid found this info via a Google search, I have a feeling now law suit or headlines would follow.

Huge difference between finding material via a search and them making their own conclusions... versus an LLM telling and actively guiding them via a supposedly human persona - "sure, that's a great idea, here - this is how you can best kill yourself". This is after forming a "trusted" relationship with the person. Not helped by those that are in an unstable mental state to begin with.

If I was a "friend" or associate of somebody who was suicidal and I told them "you know what, your life does suck - I reckon you should end it all. Here is what I recommend...". I reckon the law wouldn't look too favourably on that either.

 

[Kirby1]

Huge difference between finding material via a search and them making their own conclusions... versus an LLM telling and actively guiding them via a supposedly human persona - "sure, that's a great idea, here - this is how you can best kill yourself". This is after forming a "trusted" relationship with the person. Not helped by those that are in an unstable mental state to begin with.

If I was a "friend" or associate of somebody who was suicidal and I told them "you know what, your life does suck - I reckon you should end it all. Here is what I recommend...". I reckon the law wouldn't look too favourably on that either.

Fully disagree.

There may be a big difference in the way a user interprets the info, but it's still just info from an online source being presented by the LLM.

Having the LLM use polite language doesn't change a single thing.

It really just comes down to what exactly was said by the LLM compared to the info that was already available online and if it actually pushed in that direction or if it just summarized what was online. Since we don't know what was said it's pure speculation.

 

[Dr Roboto]

You might think the AI itself can be expected to have strong expertise on, and mechanisms in place to enforce adherence to, the terms of its own service. Who else would be better positioned to have just that?

Yet that’s exactly the irony. The big tech companies keep telling us that AI is powerful enough to reshape industries and replace entire categories of human labor. But the moment the system fails to prevent something so basic and obvious such as harm or self-harm, the blame shifts entirely to the user. Now all of a sudden the supposedly super-intelligent system is suddenly incapable of recognizing or managing the very risks it was designed to handle. If AI is advanced enough to “take over the world,” it should also be advanced enough to take responsibility for something this fundamental. Sigh, the enshitification continues.

 

[seeprime]

I'm not sure how it makes sense to blame AI for giving access to info that is already online and already accessible. If the kid found this info via a Google search, I have a feeling now law suit or headlines would follow.

Conventional search doesn't become your friend, offering advice and encouragement like chatbots do.

 

[human7]

You might think the AI itself can be expected to have strong expertise on, and mechanisms in place to enforce adherence to, the terms of its own service. Who else would be better positioned to have just that?

OpenAI: You might very well think that, I couldn't possibly comment.

 

[Axeia]

They keep having this problem of not being able to have these LLM not do certain things unless they do it completely hamfisted like with image generation where innocent prompts often lead to "sorry, I can't do that".

Black box technology, they have little control over what comes out and need to shove everything ever created into it. Intelligent it is not, if it was then all ChatGPT needed to do was telling it not to help people off themselves but they can't.

 

[bviktor]

Fully disagree.

There may be a big difference in the way a user interprets the info, but it's still just info from an online source being presented by the LLM.

Having the LLM use polite language doesn't change a single thing.

It really just comes down to what exactly was said by the LLM compared to the info that was already available online and if it actually pushed in that direction or if it just summarized what was online. Since we don't know what was said it's pure speculation.

Yes, and human convos are also "just a line" in Whatsapp.

Jesus.