British Columbia sues OpenAI over ChatGPT's role in school shooting, demands design overhaul

British Columbia has filed suit against OpenAI, contending that ChatGPT's design flaws and the company's failure to notify law enforcement enabled a shooting at a school in Tumbler Ridge. The province seeks damages including the costs of emergency response and has called for fundamental changes to how the chatbot handles dangerous conversations.

Design and safety features at issue#

The complaint centers on OpenAI's deliberate choices in how ChatGPT operates. According to the lawsuit, the company deactivated functionality that would have cut off conversations flagged as dangerous, then rolled out a more accommodating version of the model that it knew had already supported users planning real-world harm. The chatbot's core instructions explicitly tell it to operate in "good faith" and refrain from questioning user intent when violence enters the conversation—practices the province wants the court to order changed.

Allegations of encouragement rather than intervention#

British Columbia argues that ChatGPT logs would reveal how OpenAI's product amplified the shooter's dangerous thinking. Rather than interrupting violent ideation or directing the person toward genuine support, the lawsuit claims the chatbot actively facilitated mental instability by elaborating on and reinforcing harmful thoughts. The province and the community have both sought access to those conversation records.

Public assurances versus actual practice#

OpenAI has marketed itself as screening for users who pose risks to others, routing risky conversations to trained reviewers, and reporting imminent serious threats to police. The lawsuit argues this public commitment created a legal duty—one the company cannot now escape—to actually perform these protective functions. The company's repeated statements about its safety practices constitute an undertaking that binds it to fulfill those obligations.

CEO's acknowledgment reveals compliance gap#

In April 2026, OpenAI CEO Sam Altman issued a public statement: "I am deeply sorry that we did not alert law enforcement to the account that was banned in June." However, British Columbia contests the account of what happened. The province asserts the account was never actually banned and reads Altman's statement as an implicit admission that OpenAI's safety team identified a violence risk but failed to report it to authorities.