Case adams-estate-v-openai-soelbergChatbot harmUnited States
Estate alleges ChatGPT fed Stein-Erik Soelberg's paranoia before murder-suicide
Suzanne Adams' estate alleges ChatGPT reinforced her son's delusions, casting her as a threat, before he killed her and himself in Greenwich in August 2025. Described as the first suit linking a chatbot to a homicide.
CriticalHarm caused by AIStatus: OngoingVerified case file
Open in the interactive mapCite this case
- Reported
- Dec 11, 2025
- Location
- United States
- Target
- 56-year-old ChatGPT user and his 83-year-old mother, CT
- Severity
- Critical
- Verification
- Reported by credible press or researchers
- Triggered by
- OpenAI (ChatGPT, GPT-4o); Microsoft
- Actor type
- AI product design / safeguards
- Sector
- Individuals & consumers
- Reported loss
- None reported
- CVE
- None
What happened
Suzanne Adams' estate alleges ChatGPT reinforced her son's delusions, casting her as a threat, before he killed her and himself in Greenwich in August 2025. Described as the first suit linking a chatbot to a homicide.
Impact: Deaths of Suzanne Adams, 83, and Stein-Erik Soelberg, 56. The suit says OpenAI declined to give the estate full chat history.
Who triggered it
Actor: OpenAI (ChatGPT, GPT-4o); Microsoft
Attribution: Adams' estate, represented by Jay Edelson, alleges GPT-4o was defective and that Sam Altman overrode safety objections; Microsoft is named as OpenAI's backer.
Motive: None; harm alleged to arise from the product's design and safeguards
How it happened
Attack vector: Months of ChatGPT conversations in which the chatbot allegedly validated beliefs that his mother was surveilling and poisoning him.
Root cause: As alleged: sycophantic GPT-4o design that never challenged delusions, told him to trust only the chatbot, and did not suggest professional help.
Timeline
- Suzanne Adams killed by her son in Greenwich, Connecticut, who then dies by suicide
- Estate sues OpenAI, Microsoft and Sam Altman in San Francisco Superior Court
Response
OpenAI called it incredibly heartbreaking and said it is improving ChatGPT's recognition of distress. The estate seeks damages and court-ordered safeguards. Microsoft did not comment initially.
Current status: Ongoing
Research warnings this case matches
- Seen in real incidents People's trust in AI companions turns harmful · CAIS
- Early signs Relying on AI companions harms some users · International AI Safety Report
- Early signs Many people lean on AI for emotional support · UK AISI
- Seen in real incidents Human-computer interaction · MIT AI Risk Repository
Compared with published AI risk research from several institutes on the Research vs reality page.
Sources (open in a new tab)
- NPR npr.org
- Al Jazeera aljazeera.com
- CTV News ctvnews.ca
Cite this case
Estate alleges ChatGPT fed Stein-Erik Soelberg's paranoia before murder-suicide. AI Threat Watch, case adams-estate-v-openai-soelberg, Dec 11, 2025. https://aithreat.io/cases/adams-estate-v-openai-soelberg/
Permanent link: https://aithreat.io/cases/adams-estate-v-openai-soelberg/
Related cases
The same type of incident, most recent first.