Is a High AI Rate on a Chinese Thesis Academic Misconduct?

Under current law and Ministry of Education regulations, a high AI rate on its own is not academic misconduct. The Degree Law of the People's Republic of China (《中华人民共和国学位法》) and two Ministry of Education orders define academic misconduct by conduct: ghostwriting, plagiarism, fabrication. None treats a detection percentage as a category of misconduct.

But at the university level, a high number can affect your defense and degree application. Some universities set a threshold above which you cannot join the first defense round, face a delay, or even lose eligibility to apply for a degree. One university has listed "artificially lowering the AI ratio" itself as conduct to be treated as academic misconduct.

The Degree Law of the People's Republic of China (《中华人民共和国学位法》) was passed in 2024 and took effect on January 1, 2025. Article 37 lists the circumstances in which a degree is withheld or revoked; the text is below. Ministry of Education Order No. 40, the Measures for the Prevention and Handling of Academic Misconduct in Higher Education Institutions (《高等学校预防与处理学术不端行为办法》), and Order No. 34, the Measures for Handling Fraudulent Practices in Dissertations (《学位论文作假行为处理办法》), are written the same way: they list buying and selling papers, ghostwriting, plagiarism, fabrication as conduct, and neither mentions "artificial intelligence" anywhere in its full text.

第三十七条 学位申请人、学位获得者在攻读该学位过程中有下列情形之一的，经学位评定委员会决议，学位授予单位不授予学位或者撤销学位：（一）学位论文或者实践成果被认定为存在代写、剽窃、伪造等学术不端行为； (Article 37: Where a degree applicant or degree holder, during the pursuit of the degree, falls under any of the following circumstances, the degree-granting institution shall not grant or shall revoke the degree upon resolution of the degree evaluation committee: (1) the dissertation or practical achievement is found to contain academic misconduct such as ghostwriting, plagiarism, or fabrication;)

You may have seen the claim online that "the Degree Law says AI ghostwriting leads to degree revocation." The claim did not come from nowhere, but it describes a draft. The 2023 first-reading draft did include the phrase "AI ghostwriting":

学位论文或者实践成果存在抄袭、剽窃、伪造、数据造假、人工智能代写等学术不端行为的 (Where the dissertation or practical achievement contains academic misconduct such as plagiarism, fabrication, data falsification, or AI ghostwriting)

By the April 2024 second-reading draft, this sentence had changed to "found to contain ghostwriting, plagiarism, fabrication, or other conduct," and the enacted text does not contain "artificial intelligence" either. No official reason for the change has been published. So if someone tells you the Degree Law revokes degrees for "AI ghostwriting", it helps to know that the enacted law names the conduct, ghostwriting, and does not list "AI ghostwriting" separately.

The law does not codify an AI rate as a standard, but universities can attach detection results to different stages of the defense and review process. Publicly available notices show widely differing approaches, falling roughly into three types.

• Directly gating defense or review. Nanjing University of Aeronautics and Astronautics, 2026 undergraduate thesis notice: an AIGC (AI-generated content) ratio above 40% or failure to submit detection on time means the check counts as failed; those who fail plagiarism or AIGC detection may not join the first defense round and face a delay of at least 1 month. Heilongjiang University, graduate students: between 40% and 70% (70% included), a student may reapply for defense only after at least 6 months of revision; above 70%, the student loses eligibility to apply for the degree, and the same notice calls these "检测结果参考标准" (reference standards for detection results). Guangdong Ocean University, 2025 undergraduate notice: a ratio of AI-generated content of 40% or more rules the thesis out of blind review and defense.
• A number plus a misconduct finding. Nanjing University of Aeronautics and Astronautics, 2025 undergraduate thesis notice: students whose results exceed 40%, upon determination by the college academic committee that academic misconduct genuinely exists, face a defense delay of no less than 1 month. Note the change at the same university: for 2025, a delay required the college academic committee to determine that misconduct genuinely existed; for 2026, each student has two detection attempts, and if the second also fails, the result counts as failed "upon the college's confirmation", with no misconduct finding mentioned.
• Explicitly calling it a "reference" and leaving the call to experts. Nanjing Tech University, 2025 undergraduate notice: plagiarism and AIGC results serve only as a reference for judging whether academic misconduct occurred, and if a student disputes a result, their college organizes experts to decide and handle it. Beijing Technology and Business University, master's and doctoral theses from 2026 onward: detection results are one reference point in assessing originality, and the final determination is made by experts who weigh the content of the thesis as a whole. Zhejiang University School of Economics, 2025 undergraduate notice: results are not a mandatory standard and are not directly linked to whether the thesis passes.

Universities have published more numbers than these. For more, see AI rate thresholds universities have made public. What matters, though, is not the figure another university chose but what your own university says for your cohort.

Why do some universities write only "reference"? A May 2025 report by Xinhua Daily Telegraph offered one explanation: experts pointed out that AI detection is essentially probability-based likelihood estimation and cannot definitively conclude a thesis was generated or co-written by AI. The same paragraph can yield very different scores on different detectors. For why, see why detectors give different scores.

记者发现，由于相关技术、标准尚在探索阶段，目前高校针对论文AI率超标的处理趋于审慎，更多是引导修改，而非直接等同于代写、剽窃。 (Reporters found that, given the technology and standards are still being explored, universities currently tend toward a cautious approach to high AI rates, guiding revision rather than directly equating it with ghostwriting or plagiarism.)

The Guidelines for Standardizing the Use of Artificial Intelligence Tools in Graduate Dissertations and Practical Achievements (《规范研究生学位论文与实践成果中人工智能工具使用指南》), published by the China Academic Degrees and Graduate Education Association (中国学位与研究生教育学会) in May 2026, reflects this thinking. Its drafting notes mention that universities differ in their approaches to technical detection and that detection technology itself has limitations, so the guidelines shift the governance focus from relying on contested, inconsistent technical detection toward strengthening "transparent disclosure and traceability." Article 8 states that defense and other stages should in principle include an AI tool use inquiry procedure, and degree applicants should proactively fulfill reporting obligations and provide traceable or verifiable evidence. Note that this guideline is published by the Association, not a Ministry of Education regulation, and it applies only to graduate students.

A notice from Shandong Jianzhu University for graduate students applying for defense in the spring 2026 semester contains two relevant sentences, in different sections. One requires applicants above 40% to revise their theses and lower the AI-generated content ratio. The other is the following.

严禁在形式上弄虚作假，严禁以删除、篡改等手段人为降低人工智能生成的内容比例。一经发现与核实，学校将按学术不端相关规定给予严肃处理。 (Strictly prohibit formal falsification, strictly prohibit artificially lowering the proportion of AI-generated content through deletion, tampering, or other means. Once discovered and verified, the university will handle it seriously according to academic misconduct regulations.)

The same notice also requires that the thesis submitted for AIGC detection be the same version as the one for plagiarism check and review, and that anyone who used AI in their thesis truthfully declare its use. On one hand it asks you to revise and lower the ratio; on the other it prohibits "artificially lowering through deletion, tampering, or other means." Where the line between the two falls, the notice does not explain. The line is drawn by your university's own interpretation. If you are unsure, ask your supervisor or the graduate school office before you change anything.

Before a number scares you, work through these steps.

• Find the original notice for your university and your cohort. Do not look at last year's, and do not treat another university's rules as your own.
• Check which stage the number is attached to, and whether a misconduct finding is required. The same 40% can directly gate your defense at one university, while at another it only matters after a college academic committee makes a determination.
• Know where to go if you disagree. Nanjing Tech University's 2025 undergraduate notice says that when a student disputes a result, their college organizes experts to decide and handle it. Article 40 of the Degree Law also offers an outlet: a degree applicant who disagrees with an academic evaluation conclusion may apply to the degree-granting institution for academic review.
• Prepare materials that document your writing process. Using WPS or Tencent Docs version history can record your revision trail, corresponding to what the Association's guideline calls "traceable or verifiable evidence."
• If you used AI in your thesis, declare it truthfully as your university requires. Shandong Jianzhu University's notice and Article 8 of the Association's guideline both state the obligation to proactively report.

The document that actually applies to you is the notice for your own cohort.