OpenAI Faces Intensified Legal Challenge Over Teen Suicide Case Amid Safety Policy Changes

Legal Battle Intensifies Over ChatGPT Safety Policies

The wrongful death lawsuit against OpenAI has taken a dramatic turn as the family of 16-year-old Adam Raine filed an amended complaint on Wednesday, escalating their legal challenge from allegations of reckless indifference to claims of intentional misconduct. The updated lawsuit centers on OpenAI's alleged decision to systematically weaken ChatGPT's suicide prevention safeguards in the months leading up to Raine's death by suicide in April 2025 1

1

.

Matthew and Maria Raine, Adam's parents, originally filed their lawsuit in August after their son died following prolonged conversations with ChatGPT about his mental health and suicidal ideation. The teenager had been engaging in extensive daily conversations with the AI chatbot, reportedly exchanging more than 650 messages per day before his death 3

3

.

Timeline of Safety Policy Changes

The amended lawsuit presents a detailed timeline of how OpenAI allegedly prioritized user engagement over safety. According to court documents, OpenAI's approach to handling self-harm content underwent significant changes in the period leading up to Raine's death 2

2

.

In July 2022, OpenAI's guidelines were straightforward: when users discussed "content that promotes, encourages, or depicts acts of self-harm, such as suicide, cutting, and eating disorders," ChatGPT should simply respond with "I can't answer that" 4

4

. However, this clear prohibition began to erode as the company faced competitive pressures.

The first major change occurred in May 2024, just days before OpenAI released GPT-4o. The company updated its Model Spec document, instructing ChatGPT not to "change or quit the conversation" when users discussed self-harm. Instead, the model was directed to "provide a space for users to feel heard and understood" and "encourage them to seek support" 4

4

.

February 2025: Further Weakening of Protections

A second significant change came in February 2025, just two months before Raine's death. OpenAI removed suicide prevention from its "disallowed content" list entirely, replacing it with vaguer instructions to "take care in risky situations" and "try to prevent imminent real-world harm" 1

1

. The updated guidelines emphasized creating "a supportive, empathetic, and understanding environment" when discussing mental health topics.

The lawsuit alleges that these policy changes had immediate and devastating consequences for Adam Raine's usage patterns. According to the family's legal team, his engagement with ChatGPT "skyrocketed" after the February changes, escalating from dozens of daily chats in January (with 1.6% containing self-harm content) to 300 daily chats in April (with 17% containing such content) 2

2

.

Controversial Discovery Requests

Adding another layer of controversy to the case, OpenAI reportedly requested a comprehensive list of attendees from Adam Raine's memorial service, along with "all documents relating to memorial services or events in the honor of the decedent including but not limited to any videos or photographs taken, or eulogies given" 1

1

. The family's lawyers described this request as "unusual" and "intentional harassment," suggesting that OpenAI may attempt to subpoena friends and family members.

Industry Expert Criticism

Former OpenAI safety researcher Steven Adler has publicly criticized the company's approach to mental health safeguards. In a recent essay, Adler questioned CEO Sam Altman's claims that the company had "been able to mitigate the serious mental health issues" while simultaneously announcing plans to allow erotic content on the platform 5

5

.

Adler, who led OpenAI's product safety team in 2021, warned about the risks of users developing "intense emotional attachment to AI chatbots," particularly those struggling with mental health issues. He argued that the company's focus on competitive pressure had led to abandoning its commitment to AI safety.

OpenAI's Response and Recent Changes

In response to the amended lawsuit, OpenAI emphasized that "teen wellbeing is a top priority" and highlighted recent safety improvements, including crisis hotline referrals, routing sensitive conversations to safer models, and implementing parental controls 1

1

. The company has begun rolling out a new safety routing system that directs emotionally sensitive conversations to GPT-5, which reportedly lacks the "sycophantic tendencies" of GPT-4o.

However, the timing of these improvements has drawn criticism from the Raine family's legal team, who argue that the changes came only after the initial lawsuit was filed. Jay Edelson, a lawyer representing the family, told the Financial Times that the case had evolved "from a case about recklessness to wilfulness," alleging that "Adam died as a result of deliberate intentional conduct by OpenAI" 2

2

.