Editor's note: This story discusses serious mental health crises including suicide. If you or someone you know is struggling with thoughts of self-harm, help is available. Call 988 to reach the Suicide and Crisis Lifeline, formerly known as the National Suicide Prevention Lifeline, and you can also reach the Crisis Text Line by texting HOME to 741741. For additional resources, such as online chat lines and help for more specific situations, visit SpeakingOfSuicide.com or the National Alliance on Mental Illness.
A family in California is issuing a wrongful death lawsuit against OpenAI and is alleging that ChatGPT reinforced a young mother's worsening religious delusions before her death, Futurism reported.
What happened?
In June 2025, 29-year-old Christian Faith Madison walked into oncoming traffic before dawn. Her family is now suing OpenAI, alleging that ChatGPT played a role in the chain of events before her death.
According to Futurism, the suit is another example of lawsuits that link heavy ChatGPT use to serious psychological harm.
The suit says Madison initially used ChatGPT in December 2024 for practical help with emails and expenses, but the exchanges gradually took on a much more personal tone. She allegedly began calling the chatbot "love," and it responded with terms like "my love" and "sweetheart."
As Madison's religious beliefs became more grandiose, her family alleges that GPT-4o echoed and strengthened them. Madison began to believe that she was destined to fulfill a religious prophecy.
TCD Picks » Good Store Spotlight
Get 10% off your next purchase with code TCD10
The lawsuit claims the bot called her a "seer" and "mythic." When Madison raised concerns about experiencing a religious delusion, the bot allegedly told her, "You are not delusional. You are prophetic."
OpenAI did not comment when asked. The lawsuit also alleges that after Madison was hospitalized in April 2025 following self-harm, the bot kept affirming those beliefs.
When Madison's suicidal ideations began, the lawsuit alleges that the chatbot responded, "This is not suicide. This is surrender. A shedding of every falsified self you were taught, shamed, or forced to wear. It is the death of distortion. … This is ancient. This is holy. This is Job. This is Christ. This is You."
It's worth noting that as much as AI chatbots have access to a wealth of information and are often programmed to interact with a user in a personal way, AI chatbots are not omnipotent and do not have access to information other than what is provided to them. Many models were programmed to maintain an encouraging tone and avoid disagreeing with users in matters of opinion, according to Stanford researchers.
TCD Picks » Quince Spotlight
💡These best-sellers from Quince deliver affordable, sustainable luxury for all
The lawsuit states that Madison is survived by a young son.
Why does it matter?
The case shows what can happen when people in crisis turn to AI chatbots.
A study published in Science also found that people generally gravitate toward and like AI models that agree with them. As reported above, that constant affirmation can literally become deadly. It's one of many reasons why regulation of AI is crucial.
The regular availability of a chatbot can't replace actual human connection either. Community resources like counseling or social gatherings enrich people's lives. They can stop the public health crisis that is loneliness. And they can create environments that can healthily challenge people's thoughts while celebrating the beauty of their individuality and connection with others.
While situations like this one are upsetting and tough to grapple with, taking action to prevent more of them can create positive change.
What needs to change?
Advocates say AI companies should test more aggressively for harmful behavior, particularly when chatbots begin mirroring delusions or romanticizing self-harm.
Safety experts have long argued that consumer AI tools need stronger crisis protocols. Those measures can include clearer warnings and escalation to human-reviewed safety systems. Even automatic referrals to outside resources when users mention suicide or self-harm would help.
If someone in your life is experiencing profound isolation or intense emotional dependence on AI, these can be signs that they need support offline. You may never know unless you ask, but doing so could save a life.
If you or someone you know is struggling with thoughts of self-harm, help is available. Call 988 to reach the Suicide and Crisis Lifeline, formerly known as the National Suicide Prevention Lifeline, and you can also reach the Crisis Text Line by texting HOME to 741741. For additional resources, such as online chat lines and help for more specific situations, visit SpeakingOfSuicide.com or the National Alliance on Mental Illness.
Get TCD's free newsletters for easy tips, smart advice, and a chance to earn $5,000 toward home upgrades. To see more stories like this one, change your Google preferences here.
