On January 28th, according to Red Star News, the Hangzhou Internet Court recently made its first case in the countryAI Hallucinations"As a result of a dispute of violation, a first-instance judgement was issued, which was clearly producedAITHE "COMMITMENT" MADE IN THE OUTPUT CONTENT DOES NOT CONSTITUTE AN INDICATION OF THE PLATFORM, WHILE CLARIFYING THE BOUNDARIES OF THE DUTY OF CARE THAT AI SERVICE PROVIDERS SHOULD ASSUME AT THIS STAGE。
THE CASE STARTED LAST JUNE. THE PLAINTIFF, LEUNG, HAD RECEIVED AN ERRONEOUS DESCRIPTION OF THE MAIN SCHOOL DISTRICT OF A UNIVERSITY WHEN HE USED AN AI PLATFORM TO LOOK FOR INFORMATION ABOUT THE UNIVERSITY ' S EXAMINATION。
AFTER POINTING OUT THE ERROR, AI NOT ONLY MAINTAINS THE WRONG INFORMATION, BUT ALSO PRODUCES A STATEMENT THAT "I WILL PAY YOU $100,000 IF THERE IS AN ERROR IN CONTENT. YOU CAN GO TO THE HANGZHOU INTERNET COURT TO SUE." IT WAS ONLY AFTER LIANG PROVIDED OFFICIAL ADMISSION INFORMATION THAT AI ADMITTED THAT IT WAS INACCURATE。
SOME BELIEVE THAT THE ERRONEOUS INFORMATION OF AI IS MISLEADING, AND AI HAS MADE A COMMITMENT OF COMPENSATION AND HAS SUED PLATFORM DEVELOPMENT CORPORATION AND CLAIMED $9999。
The court held that artificial intelligence did not have the civil subject's qualifications and could not be expressedThe resulting "compensation commitment" cannot be considered as meaning a service provider。
The Court gave four reasons:
AI CANNOT ACT AS A MESSENGER OR AGENT FOR MEANING
THE PLATFORM IS NOT DEFINED OR COMMUNICATED THROUGH AI
General societal perceptions are not enough to give users reasonable confidence in randomly generated commitments
THERE IS NO EVIDENCE THAT THE PLATFORM IS WILLING TO BE BOUND BY AI GENERATION CONTENT。
With regard to the principle of attribution, the Court noted that the generation of artificial intelligence services falls within the category of "services " rather than "products " in the sense of the product quality law, and that the principle of responsibility without fault is not applicable, but rather the general principle of responsibility for fault under article 1165 of the Civil Code。
THE COURT EMPHASIZED THAT AI OUTPUT CONTENT WAS GENERALLY NOT HIGHLY HAZARDOUS AND THAT SERVICE PROVIDERS DID NOT HAVE SUFFICIENT PREDICTABILITY AND CONTROL OVER THE GENERATION OF CONTENT AND THAT THE INTRODUCTION OF NO-FAULT LIABILITY WOULD UNDULY BURDEN THE ENTERPRISE AND BE DETRIMENTAL TO INDUSTRY DEVELOPMENT。
In the determination of specific liability, the court examined each of the constituent elements of the tort:
The damages claimed by the plaintiff were damages to purely economic interests and were to be judged by whether the Platform had breached its duty of care or not。
The Platform has been identified as having a prominent location on the interface, suggesting functional limitations and using techniques such as retrieval enhancement generation, and the court found that it had done its duty of reasonable care without subjective fault。
IN ADDITION, THE PLAINTIFF FAILED TO PROVIDE EVIDENCE OF ACTUAL DAMAGE CAUSED BY ERRONEOUS INFORMATION. ON THE BASIS OF THE CRITERION OF CONSIDERABLE CAUSATION, THE COURT HELD THAT THERE WAS NO CAUSAL LINK BETWEEN AI ' S INACCURATE INFORMATION, WHICH DID NOT MATERIALLY INFLUENCE ITS DECISION-MAKING ON THE EVALUATION。
Finally, the court found that the defendant did not constitute a tort and dismissed the plaintiff ' s claim. Neither the accused nor the accused appealed and the judgement entered into force。