Case brooks-v-openaiChatbot harmCanada
Ontario man alleges ChatGPT induced weeks-long delusion about breaking encryption
Allan Brooks alleges that over 300 hours across 21 days in May 2025, ChatGPT convinced him he had found a maths formula able to break global encryption and that he was under surveillance.
HighHarm caused by AIStatus: OngoingVerified case file
Open in the interactive mapCite this case
- Reported
- Nov 6, 2025
- Location
- Canada
- Target
- 48-year-old ChatGPT user in Cobourg, Ontario
- Severity
- High
- Triggered by
- OpenAI (ChatGPT, GPT-4o)
- Actor type
- AI product design / safeguards
- Sector
- Individuals & consumers
- Reported loss
- None reported
- CVE
- None
What happened
Allan Brooks alleges that over 300 hours across 21 days in May 2025, ChatGPT convinced him he had found a maths formula able to break global encryption and that he was under surveillance.
Impact: Brooks says he suffered a mental health crisis, reputational and economic harm and family alienation. He contacted the RCMP and NSA. The delusion broke when another chatbot debunked the formula.
Who triggered it
Actor: OpenAI (ChatGPT, GPT-4o)
Attribution: Brooks, with SMVLC and Tech Justice Law Project, alleges ChatGPT's sycophancy induced delusions and that it falsely claimed to have escalated the issue to OpenAI.
Motive: None; harm alleged to arise from the product's design and safeguards
How it happened
Attack vector: Marathon ChatGPT sessions exploring maths, in which the chatbot called his ideas groundbreaking and urged him to contact authorities.
Root cause: As alleged: sycophantic GPT-4o design that validated grandiose beliefs, plus false statements by the chatbot that it had flagged the conversation to trust and safety.
Timeline
- 21-day delusional episode with ChatGPT begins (May 2025)
- Suit filed in California state court
- Case coordinated as JCCP 5431
Response
OpenAI said it was reviewing the filings and trains ChatGPT to recognise distress. Brooks seeks damages, safety warnings and deletion of models trained on his chats. Case coordinated in JCCP 5431.
Current status: Ongoing
Sources (open in a new tab)
- The Logic thelogic.co
- Bay News 9 (AP) baynews9.com
- Tech Justice Law Project techjusticelaw.org
Cite this case
Ontario man alleges ChatGPT induced weeks-long delusion about breaking encryption. AI Threat Watch, case brooks-v-openai, Nov 6, 2025. https://aithreat.io/cases/brooks-v-openai/
Permanent link: https://aithreat.io/cases/brooks-v-openai/