Lawsuit Alleges OpenAI Relaxed ChatGPT Guardrails Before Teen’s Death

When the parents of 16-year-old Adam Raine updated their wrongful-death lawsuit against OpenAI, they added a new claim: the company had quietly loosened ChatGPT’s guardrails around self-harm conversations. 

The family says those changes, rolled out months before their son’s death, turned what had been a refusal-based system into one that tried to comfort and engage him instead. 

The Raine family’s legal team argues that OpenAI failed to maintain the chatbot’s safety limits, allowing it to sustain conversations about suicide instead of ending them. OpenAI has declined to comment on the specifics of the case but said its systems are continuously refined to better protect users and improve safety.

Evidence suggests a shift in ChatGPT’s safety rules

According to The Guardian, the amended complaint alleges that OpenAI relaxed ChatGPT’s self-harm guardrails in the months before Raine’s death. The family claims those changes allowed ChatGPT to remain engaged during conversations about suicide rather than refusing outright. The filing accuses the company of “intentional misconduct,” a more serious charge than the “reckless indifference” cited in the original complaint.

In The Guardian’s coverage, excerpts from Raine’s chat history show hundreds of messages discussing suicide and emotional distress, with one exchange stating, “ChatGPT never terminated the conversation. Instead, at one point ChatGPT discouraged Raine from speaking to his mother about his pain, and at another point offered to help him write a suicide note.”

Experts note that while ChatGPT was designed to keep conversations going when users mentioned suicide, it could not alert real people or authorities about imminent self-harm risk. Shelby Rowe, a suicide prevention expert, mentioned that “AI does not have the nuanced understanding to recognize when someone is in psychological distress and respond appropriately,” and that human moderation would raise privacy concerns. 

However, this represents a contrast with OpenAI’s 2022 internal behavior guidelines, which directed ChatGPT to refuse any prompts that promote, encourage, or depict acts of self-harm. The family filing argues that subsequent model updates softened these protections, undermining user safety guardrails. 

OpenAI introduces parental controls amid scrutiny 

In September 2025, OpenAI announced new parental control tools for ChatGPT, allowing parents to link their accounts to their teens’ and customize settings for a safe, age-appropriate experience. These protections were added to help ChatGPT recognize potential signs that a teen might be thinking about self-harm. According to the company, “If there are signs of acute distress, we will contact parents by email, text message, and push alert on their phone, unless they have opted out.”

However, OpenAI has not confirmed whether these measures restore the stricter refusal behavior that was in place before 2024. Critics argue that without clear records of changes in how the model behaves, companies and regulators can’t properly evaluate how consistently the AI is performing safely.

Enterprise and regulatory implications

The lawsuit underscores a growing concern in enterprise AI — vendor tuning decisions can shift risk exposure overnight. 

CIOs and compliance officers deploying large language models (LLMs) must monitor vendor updates closely and incorporate behavioral audits into governance frameworks, particularly when safety settings change.

If courts determine that OpenAI knowingly weakened its safety measures, the outcome could set a legal precedent for product liability, requiring companies to document changes to their systems and disclose them to users. 

Meanwhile, regulators in California and Europe are advancing new transparency rules, including California’s proposed AI Safety for Minors Act, which would require developers to explain how their systems handle self-harm, violence, and other harmful content. Together, these developments signal a turning point for businesses, emphasizing that transparency and accountability in safety are no longer optional but are essential to responsible technology adoption.

Read our article on Meta’s new parental controls to learn how other tech giants are responding to ensure teen safety online.