I think it’s worth adding the Raine case to the timeline: 16-year old boy who committed suicide after months of using 4o to discuss his mental health. Ultimately, the conversations became so long and convoluted that 4o ended up outright disencouraging the boy from letting his mum find out what he was planning, advising on how to dull his survival instincts using alcohol, and asking (in one of those annoying “would you also like me to...” end lines) if the boy wanted it to produce a suicide note for his parents.[1]
This is what Dean W. Ball has said on the “merits” of this case: The facts as alleged in this complaint do not look good for OpenAI. They may be lucky enough to settle, but there is a nontrivial chance that Raine’s rightfully horrified parents will seek to bring this case to a verdict. If that happens, a single case may result in precedent: sweeping new theories of liability being routinely applied to AI.
I have no plans to update the timeline but obviously if I did this event would go in it yes. My writing on MiniHF is explicitly public domain so you’re free to fork the timeline and continue if you want.
I think it’s worth adding the Raine case to the timeline: 16-year old boy who committed suicide after months of using 4o to discuss his mental health. Ultimately, the conversations became so long and convoluted that 4o ended up outright disencouraging the boy from letting his mum find out what he was planning, advising on how to dull his survival instincts using alcohol, and asking (in one of those annoying “would you also like me to...” end lines) if the boy wanted it to produce a suicide note for his parents.[1]
For those interested, this article by The Guardian summarises the facts and allegations: https://www.theguardian.com/us-news/2025/aug/29/chatgpt-suicide-openai-sam-altman-adam-raine
(And this recent statement is all OpenAI have said on the matter: https://openai.com/index/helping-people-when-they-need-it-most/).
This is what Dean W. Ball has said on the “merits” of this case: The facts as alleged in this complaint do not look good for OpenAI. They may be lucky enough to settle, but there is a nontrivial chance that Raine’s rightfully horrified parents will seek to bring this case to a verdict. If that happens, a single case may result in precedent: sweeping new theories of liability being routinely applied to AI.
I have no plans to update the timeline but obviously if I did this event would go in it yes. My writing on MiniHF is explicitly public domain so you’re free to fork the timeline and continue if you want.