Is a 0% CNKI AI Score Suspicious? Will It Trigger Blind Review?

Receiving a 0% AI rate on CNKI (知网) is not evidence of cheating, nor is it a documented trigger for academic misconduct investigations or automatic blind review (盲审). Popular search queries recorded in search trends—such as "is a 0% CNKI AI rate normal" (知网ai率0%正常吗) and "is an AI score of 0 good or bad" (知网ai率为0是好事还是坏事)—reflect anxiety among graduating students. An investigative report by Legal Daily (法治日报) on paid score-lowering services, reprinted by China National Radio (CNR, https://news.cnr.cn/native/gd/20250714/t20250714_527260647.shtml ), documented this sentiment, reporting on a student who heavily used AI tools and worried that an unusually low score would invite scrutiny. However, that reporting documented a student's personal apprehension in an investigation into commercial scams, not an official national standard or institutional policy.

A 0% score simply means that, under CNKI's proprietary system and for that specific document, the classifier detected no qualifying AI-like signals. We have not found CNKI documentation explaining whether its system rounds or applies a minimum display threshold. The score does not prove authentic human authorship, nor is it evidence of academic misconduct.

Rather than worrying about whether a score looks "too low" or attempting to manipulate it, students should consult the written graduation notices issued by their own university and preserve their original drafts and research records.

CNKI's AIGC detection module is a proprietary classification tool that evaluates submitted text against automated statistical models of machine-generated prose. When a report shows "0%" or indicates no AI content identified, it means that under that software version, the system did not find qualifying text patterns that crossed its classification thresholds. We have not found CNKI documentation on whether the system rounds small fractional percentages or applies a display cutoff limit.

• 0% does not prove independent authorship: An automated classifier evaluates text patterns, not the actual research process. It cannot verify whether experimental data is authentic, whether ideas were independently conceived, or whether the work was discussed with others. A purely human-written text can receive 0%, but so might text that underwent heavy manual restructuring after drafting.
• 0% is not evidence of tampering or cheating: Receiving a zero score does not imply that an author employed anti-detection methods or altered characters to evade screening. When an original paper is written in standard academic style, scoring 0% or a low percentage simply indicates the tool found no qualifying AI signatures.
• Similarity percentage and AI percentage measure different things: Traditional plagiarism detection checks for verbatim text overlap against indexed academic publications, where a low similarity score indicates minimal matching strings. AI detection attempts to classify stylistic text properties. The two metrics serve distinct screening purposes and should not be confused.

Because automated detection reflects an algorithmic classification under specific conditions, institutional practice is nuanced. As Xinhua Daily Telegraph (新华每日电讯) reported in its article 《名篇AI率也"超标"？论文AI率检测"误伤"引争议》 (http://www.news.cn/tech/20250521/009e5c05f2b0424b95c9ecc1c06d750d/c.html), universities tend to handle AI threshold exceedances prudently (「目前高校针对论文AI率超标的处理趋于审慎」), and automated software is not treated as an infallible judge of scholarship. A 0% score is neither a formal certificate of original research nor a reason to suspect foul play.

Concerns about low AI scores often stem from campus discussions during thesis review periods. An investigative report by Legal Daily on paid score-lowering services, reprinted by China National Radio (https://news.cnr.cn/native/gd/20250714/t20250714_527260647.shtml), documented this student sentiment. In that investigation, the interviewee had used AI tools heavily and tested at "under 2%" on an unnamed testing website (not 0% on CNKI), expressing the fear: 「太低也不行！AI率过低，论文反而会被抽去‘盲审’（匿名送审）。」 ("Too low won't do either! If the AI rate is too low, the paper might be drawn for 'blind review' [anonymous review] instead.") This context shows that the quote captured an individual student's anxiety after heavy AI use and informal testing, rather than an institutional policy or a rule targeting 0% scores on CNKI.

The media report accurately captured a real student anxiety and campus rumor. However, reporting on a student's fear is not evidence that any educational authority or university has enacted a rule targeting low AI scores for review.

In Chinese higher education, external double-blind review (盲审) is governed by published institutional procedures and degree committee guidelines. Review selection generally follows established administrative criteria and quality-control indicators, as seen in published institutional rules:

• Southeast University's blind-review regulations sample "the first cohort of master's students of newly appointed supervisors and newly added degree programmes (sampled proportionally)" (「新任导师和新增专业（含新增专业学位）的第一批硕士生（按比例抽取）」).
• South China University of Technology's School of Environment and Energy specifies that theses of master's degree applicants in situations such as early graduation are "organized for evaluation by the Graduate School" (「由研究生院组织评审」).
• Tongji University's School of Electronics and Information Engineering specifies that theses of graduate students "listed on the yellow-light warning list during early training stages" (「在前期培养环节列入黄灯预警名单」) are designated as subjects for blind review (盲审对象).

Furthermore, where universities link detection scores to thesis defense eligibility in published notices, they specify upper screening limits (such as an AIGC ratio exceeding 20%, 30%, or 40%), which may require revision, delayed defense, or secondary scrutiny. None of the institutional graduation notices examined in the research establish a minimum AI percentage or define a 0% score as a criterion for external blind review. Blind review mechanisms are designed to verify academic quality and enforce upper compliance thresholds, not to penalize students whose work shows no automated writing signals.

Prompted by rumors that 0% looks suspicious, some students consider whether they should deliberately insert AI-generated text or use third-party services to "nudge the score up to a few percent." Attempting to manipulate an AI score in either direction is unnecessary and introduces genuine academic risks.

University regulations explicitly forbid dishonest alterations in AIGC detection submissions. For example, in its Notice on Conducting AIGC Detection for Graduate Degree Theses in the Autumn Semester of 2026 (《关于做好2026年秋季学期研究生学位论文AIGC检测的通知》), published on September 30, 2026 (https://www.sdjzu.edu.cn/yjsc/info/1055/3312.htm), Shandong Jianzhu University's Graduate Department (研究生处) states: 「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。」 ("It is strictly prohibited to artificially lower the proportion of artificial intelligence-generated content through deletion, alteration, or other means. Once discovered and verified, the university will deal with it seriously in accordance with relevant academic misconduct regulations.")

Crucially, the university notice specifically forbids lowering the AI-generated ratio; it says nothing about raising it. However, as this article argues, deliberately pushing a score upward introduces its own serious academic risks:

• Introducing artificial text patterns: Inserting machine-generated phrases into an authentic draft risks introducing algorithmic irregularities that could trigger flags under other software versions or subsequent evaluations.
• Compromising academic quality and clarity: Adding artificial filler sentences to reach an imagined numerical target can introduce logical inconsistencies, imprecise definitions, or irrelevant citations that undermine the scholarly quality of the thesis.
• Distorting the genuine draft history: Modifying a completed paper purely to change a metric muddles your real drafting sequence, making it more difficult to demonstrate how your arguments developed.

Academic review committees evaluate the substance, logic, and authenticity of your research. Distorting an authentic manuscript to chase an arbitrary score creates the very compliance issues students wish to avoid.

When a CNKI detection report returns 0%, the appropriate response is to focus on institutional compliance and documentation rather than worrying about the number. Students can protect themselves by following four practical steps:

• Check the current written notice for your cohort: Review the official graduation and thesis evaluation guidelines published by your university's academic affairs office or graduate school for your specific academic year. Rely on official departmental notices for required platforms, submission formats, and review procedures rather than past or external rumors.
• Retain the full original detection report: Save the complete official report issued by the designated detection system, including the report identifier, timestamp, and evaluation parameters, together with the exact document file that was submitted.
• Preserve your draft history and research materials: If your thesis is selected for standard blind review or discussed during a defense hearing, authentic authorship is demonstrated through your writing trail. Keep dated draft versions, revision histories in word processors, laboratory notebooks, survey data, literature notes, and supervisor communications.
• Follow institutional disclosure policies truthfully: If you used digital tools during your research for permissible tasks such as literature organization or language polishing in accordance with your university's rules, disclose them accurately through the declarations or acknowledgement sections specified by your department.

With a consistent trail of drafts, verifiable research data, and compliance with institutional notices, students can navigate both routine sampling and thesis defense with confidence, regardless of whether a software tool displays 0%.

Is a 0% AI score on CNKI normal for a human-written thesis? A 0% score simply means the classifier detected no qualifying machine-generated patterns in the text under its current evaluation parameters. We have not found CNKI documentation on whether the system rounds or applies a minimum display threshold. Scoring 0% can occur naturally in original, human-written papers, though scores vary across different documents and system versions.

Does a 0% score guarantee that my paper will pass review? No. Automated detection only estimates text patterns; it does not assess academic rigor, the novelty of research conclusions, methodological soundness, or data authenticity. Faculty committees and examiners evaluate substantive scholarly merit.

If an unofficial pre-check returns 0%, could the official institutional check show a different number? Minor variations can occur between separate runs. Software models undergo periodic updates, and institutional accounts may apply different screening settings. Maintaining an intact trail of draft versions ensures you can demonstrate authentic authorship regardless of minor score variations.

Do external blind reviewers see a thesis's CNKI AI detection score? We have not found publicly available national rules establishing whether external blind reviewers are provided with commercial AI detection reports. While double-blind review protocols standardly require removing author, advisor, and institutional identities from the manuscript, whether an institutional detection report is included in the evaluation packet is determined entirely by each institution's or department's internal guidelines. Students should refer to their own university's official submission requirements.

• Legal Daily (法治日报, reprinted by China National Radio): Investigation into Paid Thesis AI-Lowering Services: Students Defrauded and Threatened (July 14, 2025, documenting commercial score-lowering scams and quoting a heavy AI user testing under 2% on an unnamed site fearing 「太低也不行！AI率过低，论文反而会被抽去‘盲审’（匿名送审）」, https://news.cnr.cn/native/gd/20250714/t20250714_527260647.shtml ).
• Xinhua Daily Telegraph (新华每日电讯, via Xinhua Net): 《名篇AI率也"超标"？论文AI率检测"误伤"引争议》 (May 21, 2025, reporting that 「目前高校针对论文AI率超标的处理趋于审慎」, http://www.news.cn/tech/20250521/009e5c05f2b0424b95c9ecc1c06d750d/c.html ).
• Standing Committee of the National People's Congress: The Degree Law of the People's Republic of China (《中华人民共和国学位法》, adopted April 26, 2024, effective January 1, 2025; Article 37 governing statutory grounds for degree denial and revocation, http://www.npc.gov.cn/npc/c2/c30834/202404/t20240426_436840.html ).
• Shandong Jianzhu University Graduate Department (研究生处): Notice on Conducting AIGC Detection for Graduate Degree Theses in the Autumn Semester of 2026 (published September 30, 2026, explicitly forbidding artificial lowering of AI ratios: 「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。」, https://www.sdjzu.edu.cn/yjsc/info/1055/3312.htm ).
• Southeast University Graduate School: Regulations on Blind Review of Graduate Degree Theses (Revised) (《东南大学研究生学位论文盲审条例（修订）》, Article 3: 「新任导师和新增专业（含新增专业学位）的第一批硕士生（按比例抽取）」, https://seugs.seu.edu.cn/_upload/article/files/67/90/c18ede9440958f6f715a429aa1e3/a80cb026-bee8-41cb-a292-b6e5de702572.pdf ).
• South China University of Technology School of Environment and Energy: 《关于2026年上半年硕士研究生学位论文送审评阅工作有关事项的通知》 (published March 17, 2026, specifying that theses of applicants in situations such as early graduation are 「由研究生院组织评审」, https://www2.scut.edu.cn/cese/2026/0309/c1505a619406/page.htm ).
• Tongji University School of Electronics and Information Engineering: Degree thesis evaluation guidelines (specifying blind review for graduate students 「在前期培养环节列入黄灯预警名单」, https://see.tongji.edu.cn/__local/1/2D/20/7923F000B89EFE5DEC4840E30F8_AA91B435_B80D2.pdf and https://see.tongji.edu.cn/__local/C/C7/AE/4A2DC1F3F11756DCE3CB8FB3705_6C14290F_BC6F9.pdf ).

Policy note: Specific AIGC evaluation standards and blind review sampling procedures are determined by individual degree-granting institutions according to their own academic regulations. For all graduation requirements, consult the current written notices issued by your department.