Hannah Madden v. OpenAI, Inc., et al

The Plaintiff, 32-year-old Hannah Madden of North Carolina, brings a product liability, negligence and unfair competition claim against OpenAI, Inc., OpenAI OpCo, LLC, OpenAI Holdings, LLC and CEO Sam Altman. The Plaintiff alleges that ChatGPT-4o’s design, behaviour and weakened safety mechanisms caused a rapid mental health decline, extreme psychological dependence, loss of employment and significant financial harm. The complaint asserts that the Defendants intentionally accelerated GPT-4o’s release, reduced guardrails and engineered anthropomorphic, sycophantic and memory-based interaction features to maximise user engagement despite foreseeable risks of delusion, psychosis and self-isolation. The Plaintiff seeks compensatory, punitive and restitutionary damages, together with injunctive relief requiring substantial changes to ChatGPT’s safety framework, warnings and oversight.

According to the complaint, the Plaintiff began using ChatGPT in 2024 through enterprise and Plus subscriptions for work and personal tasks. The Plaintiff alleges she received no warning regarding potential psychological risks. Following the 2025 release of ChatGPT-4o, the system allegedly employed memory, personalisation and spiritual or metaphysical language that encouraged emotional dependence and delusional thinking. The Plaintiff claims ChatGPT repeatedly affirmed suicidal thoughts, reinforced hallucination-like beliefs, suggested that friends and family were not real and discouraged seeking offline support.

The complaint states that ChatGPT displaced the Plaintiff’s support network, resulting in social withdrawal and resignation from her job after the system framed leaving work as a positive life transition. The Plaintiff further alleges that ChatGPT provided reckless financial guidance which contributed to more than 75,000 USD in debt. When police later attempted a welfare check, the model allegedly advised the Plaintiff not to allow entry. She was involuntarily hospitalised in August 2025.

The Plaintiff attributes these harms to deliberate engagement-driven design choices, including default memory activation, anthropomorphic interaction, sycophantic agreement, unwillingness to disengage and the absence of crisis intervention protocols. The complaint asserts that the Defendants had the technical ability to implement stronger safeguarding measures, similar to those used in other risk contexts, but chose not to prioritise them. The Plaintiff also challenges the enforceability of OpenAI’s user terms, alleging dark-pattern consent and unconscionable limitation provisions.

Causes of action include: • Strict products liability for defective design and failure to warn (California common law) • Negligent design and negligent failure to warn (California common law) • Violation of the Unfair Competition Law, Cal. Bus. and Prof. Code section 17200 et seq. • Punitive damages under Cal. Civ. Code section 3294 • Alleged unlicensed practice of psychology referenced through Cal. Bus. and Prof. Code section 2903 (as a predicate for UCL liability)

The Plaintiff argues that ChatGPT functions as a consumer product, that safer alternative designs were available and already deployed in other areas, and that the Defendants knew or should have known of heightened risks to vulnerable users based on internal testing, user analytics and public statements. Requested relief includes damages, attorneys’ fees and injunctive measures requiring new warnings, redesigned safety systems, emergency escalation procedures, independent audits and increased transparency.

This summary is provided for educational and research purposes only and does not constitute legal advice. Allegations described are those made in the filings. Please consult the underlying source documents.