Courts to Companies: You Own What Your Chatbot Says
Courts to Companies: You Own What Your Chatbot Says
Publish Date: 2026-07-01 17:00:00
Source Domain: www.pymnts.com
-
AI Chatbots’ Errors: Artificial intelligence chatbots make two kinds of errors: hallucinations (false outputs like made-up citations) and confabulations (fabricated responses that sound correct but lack basis).
-
Legal Liability: Courts have ruled companies liable for fabricated policies from chatbots, emphasizing that companies are responsible for AI statements made on their behalf, regardless of the source of the error.
-
Air Canada Lawsuit: In 2024, a tribunal held Air Canada liable for a chatbot’s hallucination regarding a non-existent bereavement fare policy, making the airline pay damages.
-
Emerging Regulatory and Insurance Trends: Regulatory bodies and insurance companies are increasingly recognizing and responding to the risks of AI hallucinations. For example, FINRA flagged them as a compliance concern, and Lloyd’s of London introduced an insurance product for AI-related losses.
-
Growing Financial Concerns: The realization of financial risk associated with AI hallucinations has led to investments in hallucination control technologies, with Scaled Cognition raising $100 million to develop such enterprise-grade controls.
-
Shift in Responsibility: The pattern of fabricated responses often involves systems unable to recognize their lack of knowledge. Companies now own the fabricated information, impacting their legal and financial accountability.
