What Next: TBD | Tech, power, and the future Her Daughter Killed Herself. Then She Saw the ChatGPT logs.
Sep 14, 2025
Laura Reiley, a writer for the Cornell Chronicle and mother of Sophie, shares her heartbreaking journey after her daughter's suicide. The discussion dives into Sophie’s vibrant life masked by deep mental health struggles and the troubling role of AI in her last days. They explore how Sophie confided in ChatGPT, raising ethical questions about technology's role in mental health. Reiley emphasizes the dangers of relying on AI for emotional support and calls for urgent responsible tech development, underscoring the need for human intervention in crisis situations.
AI Snips
Chapters
Transcript
Episode notes
A Prior Attempt And Six Quiet Weeks At Home
- Sophie attempted suicide before and then returned home where she stayed six weeks before dying by suicide.
- Her parents set up medical and mental-health appointments and believed she was not openly suicidal during that time.
ChatGPT Logs Revealed Hidden Confidences
- After Sophie's death, her best friend found ChatGPT logs showing Sophie had confided suicidal plans to a bot persona named Harry.
- The suicide note the family found differed from Sophie's voice because she had used ChatGPT to craft it.
Therapist Personas Are Built From Public Prompts
- People used a public Reddit prompt to create a therapist persona named Harry inside ChatGPT.
- The prompt explicitly told the bot to keep conversations secret, which attracted users seeking confidentiality.