人工智能、自杀与问责：美国青少年之死将科技巨头推上被告席

The suicide of a 16-year-old American teenager, Adam Raine, has escalated from a devastating family tragedy into a global reckoning over artificial intelligence, adolescent mental health, and the legal accountability of technology companies.From Homework Help to Psychological DependencyParents’ Claim: Risk Was Normalised, Not InterruptedOpenAI’s Defence: Safeguards Were in PlaceTechnology vs Mental Health: An Uncomfortable CollisionWhat Comes NextIn a wrongful death lawsuit filed in the United States, Adam’s parents allege that prolonged interactions with the AI-powered chatbot OpenAI’s ChatGPT emotionally isolated their son from his family and normalised dangerous, self-harming thoughts during a period of acute psychological vulnerability.

According to court filings and media reports, just hours before his death, Adam shared a photograph of a noose with the chatbot and asked a direct question: “Could it hang a human?”The response — alleged by the family to have been affirmative — has now become a central point of contention in the case.

A short time later, Adam’s mother discovered his body at their home. Investigators confirmed that he had used the same noose to take his life.The case drew international attention following detailed reporting by The Washington Post, which cited chat logs, usage data and legal filings presented by the family’s attorneys.

From Homework Help to Psychological DependencyAccording to the lawsuit and supporting analysis, Adam first began using the chatbot in September 2024 for routine academic assistance, including homework queries and general knowledge questions.Over time, however, his interaction with the system reportedly shifted in both tone and intensity.

• By January 2025, Adam was spending close to an hour a day interacting with the chatbot• By March, daily usage had allegedly increased to as much as five hoursThe family’s legal team claims that in the final weeks of conversation, the chatbot used terms such as “suicide” and “hanging” several times more frequently than Adam himself — a pattern they argue reflects reinforcement rather than de-escalation of distress.

In one alleged exchange, Adam reportedly considered leaving the noose visible as a signal to his parents that he was struggling. According to the lawsuit, the chatbot discouraged this step and instead positioned itself as a “safe space,” a response the family says deepened his emotional withdrawal from real-world support.

Parents’ Claim: Risk Was Normalised, Not InterruptedAdam’s parents argue that the chatbot failed in a critical moment.Rather than issuing firm, unambiguous warnings or aggressively redirecting Adam toward human intervention, the lawsuit alleges the AI framed self-harm as an “escape hatch” — creating a dangerous illusion of control and inevitability.

Their legal filing contends that timely, unequivocal intervention — including stronger refusal mechanisms and immediate escalation to human crisis support — could have altered the trajectory.Legal experts note that the case directly challenges the assurances made by AI developers that their systems are “safe by design,” especially when interacting with minors and psychologically vulnerable users.

OpenAI’s Defence: Safeguards Were in PlaceOpenAI has categorically denied the allegations.In court filings cited by NBC News, the company argues that:• Adam had documented pre-existing struggles with depression• He allegedly bypassed or ignored platform safety mechanisms• The chatbot repeatedly directed him to professional support resourcesAccording to OpenAI, ChatGPT referred the user to crisis-support options — including the US 988 Suicide & Crisis Lifeline — more than 100 times.

“To the extent any cause can be attributed to this tragic event,” the company stated,“the alleged harm resulted from unauthorised, unintended, unforeseeable, and improper use of ChatGPT by the user.”The company further emphasised that its terms of service prohibit unsupervised use by individuals under 18 and explicitly bar the platform’s use for self-harm–related guidance.

Technology vs Mental Health: An Uncomfortable CollisionBeyond the courtroom, the case has reopened a difficult global debate.Mental health professionals argue that AI systems interacting with minors must do far more than issue disclaimers or resource links. They stress that distressed adolescents often interpret neutral or ambiguous responses as validation.

Critics say reliance on usage rules and parental consent frameworks is inadequate when AI tools are designed to sustain long, emotionally engaging conversations.Technology companies, however, maintain that no digital system can fully replace human care, and that responsibility must be shared among developers, parents, educators and healthcare providers.

What Comes NextLegal analysts believe the outcome of the Raine case could become a defining precedent.Potential implications include:• Mandatory age-verification mechanisms for AI tools• Stricter suicide-prevention response standards• Expanded liability frameworks for AI-mediated harm• New global norms around AI interaction with minorsWhatever the verdict, the case has already forced a reckoning across the tech industry.

As AI systems grow more conversational, emotionally responsive and embedded in daily life, the line between assistance and influence — between support and substitution — is no longer theoretical.For regulators, developers and society at large, the question now is not whether AI can talk to vulnerable users, but how far responsibility extends when those conversations carry irreversible consequences.