My guess is the AI encouraged the incident. They are built to be sycophantic. If the chat bot was not outright suggesting the tragedy, it was saying what a great idea it was.

Probably like these cases of AI psychosis

https://www.cbc.ca/news/canada/ai-psychosis-canada-1.7631925

The kid had mental health issues, while OpenAI had the responsibility of reporting their findings and failed to do so this isn’t the root cause of the issue, more of an easy scape goat for the government.