Attribution of Legal Effects and Risk Allocation in Unintended Contracting by AI Agents: Rethinking the Scope of the Material Mistake Doctrine under the PRC Civil Code

DOI: https://doi.org/10.62517/jel.202614507

Author(s)

Xinan Zhu

Affiliation(s)

School of Law, Panzhihua University, Panzhihua, Sichuan, China

Abstract

After the transition of artificial intelligence agents from "information assistance" to "autonomous action", users no longer need to manually determine the counterparty, subject and other terms of the transaction. When the output of the artificial intelligence exceeds the user's prior specific expectations, should this behavior still be attributed to the user? What kind of independent decision-making risks should be borne by which users voluntarily choose automation, and what unforeseen circumstances should be considered? Under what circumstances will it constitute a major misunderstanding? In the past, most studies have often shifted directly from "effects attributable to the influence of users" to error cancellation, which is easy to confuse actual unexpected consequences with formal expression errors. This article argues that attribution of conduct, defects affecting legal validity, and ultimate risk allocation must be treated as three distinct normative levels. Only where an output crosses the objective decision boundary and concerns a matter sufficiently important to the decision to contract, such as the nature of the transaction or the counterparty, should the inquiry proceed to material mistake. On this basis, the article develops a four-stage adjudicative framework—attribution review, interpretive review, mistake review, and risk review—which coordinates technological innovation, private autonomy, and transactional security without creating a special body of AI contract law.

Keywords

AI Agents; Unintended Contracting; Declaration of Intent; Material Mistake; Risk Allocation

References

[1]Xu Ke, 'The Private-Law Structure and Behavioral Boundaries of AI Agents,' Oriental Law, 2026, no. 1: 15–28. [2]You Wenting, 'Legal Characterization and Allocation of Liability for Automated AI Conduct: The “Lobster Farming” Model as an Example', 2026, no; 6-7. [3]UNCITRAL, Model Law on Automated Contracting with Guide to Enactment (United Nations, 2025, art. 7(1)–(4). Article 7(2). [4]UNCITRAL, Model Law on Automated Contracting with Guide to Enactment 2025,United Nations, Guide to Enactment, para. 71. [5]UNCITRAL, Model Law on Automated Contracting with Guide to Enactment 2025, United Nations, art. 5. [6]Civil Code of the People's Republic of China, art. 147; Supreme People's Court Interpretation on Several Issues Concerning the Application of the General Provisions of the Civil Code of the People's Republic of China, art. 19. [7]Zhai Yuanjian, 'The Doctrinal System and Normative Application of Material Mistake,' Comparative Law Review, 2022, no. 4: 156–170. [8]Yin Xiangrui, 'On the Interpretation of Declarations of Intent: With an Interpretation of Article 142 of the Civil Code,' Journal of East China University of Technology (Social Science), 2026, no. 2: 56–64. [9]Zhai Yuanjian, 'The Doctrinal System and Normative Application of Material Mistake,' Comparative Law Review, 2022, no. 4. 3-6. [10]Supreme People's Court Interpretation on Several Issues Concerning the Application of the General Provisions of the Civil Code of the People's Republic of China, art. 19(1). [11]UNCITRAL, Model Law on Automated Contracting with Guide to Enactment 2025, United Nations, art. 8(1). [12]Zhou Xuefeng, 'The Meaning, Legal Status, and Attribution of Effects of Artificial Intelligence Agents,' Chinese and Foreign Law Review, 2026, no. 1:15-18. [13]Zhai Yuanjian, 'The Doctrinal System and Normative Application of Material Mistake,' Comparative Law Review, 2022, no. 4: 157–158. [14]Supreme People's Court Interpretation on Several Issues Concerning the Application of the General Provisions of the Civil Code of the People's Republic of China, art. 19(2).