Typological Determination of the Negligence Liability of Generative Artificial Intelligence Providers
DOI:
https://doi.org/10.54097/weex6x65Keywords:
Generative artificial intelligence; providers; negligent crimes; typological determination.Abstract
The iterative evolution of generative artificial intelligence has given rise to novel criminal risks. For a long time, academic debate has remained locked in the binary controversy over whether strong artificial intelligence qualifies as a subject of criminal liability, while few studies have turned to the practical question of how liability should be attributed to providers. Attribution thinking remains trapped within the conventional paradigm of subject-status determination. Provider liability is often presumed to be liability for negligent crimes, with accountability pursued by stretching the causal chain, yet the preconditions for establishing negligence liability have not been adequately examined, the boundaries of attribution remain vague, and no systematic determination framework has been constructed. Moving beyond the controversy over subject status, this paper divides scenarios according to differences in the logic of harm production, and the paths of attribution diverge accordingly. In traditional negligence scenarios, harm mostly stems from human carelessness at the source stage, and autonomous algorithmic decision-making does not substantially intervene in the causal process; the traditional logic of negligent crimes can therefore be applied by relying on foreseeability, the duty of result avoidance, and adequate causation. In scenarios of independent harm infliction by artificial intelligence, by contrast, autonomous algorithmic output breaks the causal chain, the theory of supervisory negligence faces dogmatic difficulties, and the focus of attribution shifts to the post-hoc removal duty, so that legal interests are protected through the path of duty offenses. Between the protection of legal interests and technological innovation, the criminal law should maintain modesty and treat administrative regulation as the preferred path of governance.
Downloads
References
[1] Xi Jinping: ‘Working Together to Build a Fair and Reasonable Global Artificial Intelligence Governance System’—Keynote Speech at the Opening Ceremony of the 2026 World Conference on Artificial Intelligence and the High-Level Conference on Global AI Governance, published in the Gazette of the State Council, July 17, 2026, No. 21 of 2026.
[2] Sha Rina. Determination of the World’s First Fatal Autonomous-Driving Accident: Uber Not Criminally Liable. People’s Daily Online. March 7, 2019. Information on http: // auto. people.com.cn/n1/2019/0307/c1005-30962032.html (last visited August 26, 2026).
[3] Tesla Held Partly Liable in the Fatal Autopilot Crash Case. Tencent News. September 16, 2025. Information on: https://news.qq.com/rain/a/20250916A058MF00 (last visited August 26, 2026).
[4] Liu Xianquan. A Re-Examination of Theoretical Research on the Evolution of the Subject of Criminal Liability in the Artificial Intelligence Era. Law Science. No. 9 (2025), p. 82-94.
[5] Ren Yanjun, Nie Ran. The Negation of the Subject Qualification of Strong Artificial Intelligence for Criminal Liability. Tribune of Social Sciences. No. 4 (2024), p. 132-141.
[6] ProPublica. Machine Bias: There’s Software Used Across the Country to Predict Future Criminals. And It’s Biased Against Blacks. ProPublica. 23 May 2016. Information on: https: //www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing.
[7] Zhang Xuehong. Final Judgment in Beijing’s First Criminal Case of Damaging an Artificial Intelligence Model: Engineer Convicted of the Crime of Damaging Computer Information Systems and Sentenced. Guangming Online. August 19, 2026. Information on: https: // www.toutiao.com/article/7675557807453651506/?wid=1787113442186 (last visited August 27, 2026).
[8] Interim Measures for the Management of Generative Artificial Intelligence Services, Article 14: ‘Where a provider discovers unlawful content, it shall promptly take disposal measures such as stopping generation, stopping transmission, and elimination, take measures such as model optimization and retraining to rectify the situation, and report to the relevant competent department.’
[9] Ulrich Beck. Risk Society. Trans. He Bowen. Yilin Press, 2004, p. 66.
[10] Zheng Zexing, Zeng Qinwen. On the Criminal Liability of Operators of Generative Artificial Intelligence Services. Contemporary Law. No. 5 (2025), p. 52-62.
[11] Chu Chencheng. Indirect Governance: The Basic Model of Criminal Law’s Response to Digital Technology Risks. Legal Forum. No. 1 (2025), p. 65-77.
[12] Fang Huiying. Criminal Risks and Prevention Strategies of Generative Artificial Intelligence—Taking ChatGPT as an Example. Journal of Nanchang University (Humanities and Social Sciences Edition). No. 4 (2023), p. 52.
[13] Claus Roxin. German Criminal Law (Vol. 1). Trans. Wang Shizhou. Law Press, 2005, p. 392.
[14] Colin Scott. Regulation, Governance and Law: Research on Frontier Issues. Trans. An Yongkang. Tsinghua University Press, 2018, p. 290-291.
[15] Zhang Weiyao. Result Attribution in Negligent Crimes of Generative AI Producers. Journal of East China University of Political Science and Law. No. 6 (2025), p. 75-85.
[16] Wang Chunyu. The Determination and Regulation of Criminal Liability for Crime Methods Output by Large Models. Jiangxi Social Sciences. No. 6 (2026), p. 197-205.
[17] Fabio Basile, Beatrice Fragasso. Intelligenza artificiale e diritto penale: quattro possibili percorsi di indagine. Diritto Penale e Uomo. 2019, p. 1-33. Information on https: // www.penalecontemporaneo.it/upload/3089-basile2019.pdf.
[18] Gabriel Hallevy. The Basic Models of Criminal Liability of AI Systems and Outer Circles. SSRN Electronic Journal. (11 June 2019), p. 1-16. Information on: https://ssrn.com/abstract=3402527. DOI: https://doi.org/10.2139/ssrn.3402527
[19] Wang Shan, Wang Jing. Attribution of Responsibility for the Abuse of Artificial Intelligence Technology and the Criminal Law Response. DeHeng Law Offices website. March 6, 2026. Information on: https://www.deheng.com/content/26410.html (last visited September 25, 2026).
[20] EU Artificial Intelligence Act, Article 99(3): ‘Non-compliance with the prohibited AI practices referred to in Article 5 shall be subject to administrative fines of up to 35,000,000 euros; where the offender is an enterprise, the fine may be up to 7% of its total worldwide annual turnover in the preceding financial year, whichever is higher.’
[21] Editorial Board of China Procuratorial Yearbook (ed.). China Procuratorial Yearbook (2024 Volume). China Procuratorial Press, 2024, Part VI ‘Statistical Data.’
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Academic Journal of Management and Social Sciences

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

