Are Thesis AI-Lowering Services Reliable? How to Vet Third-Party Intermediaries

In investigative reports published by major media outlets, commercial intermediaries that claim to "lower thesis AI rates" (代降 AI 率) are overwhelmingly unreliable, operate in an unregulated grey market, and expose students to severe academic, legal, and personal security risks.

Public investigations conducted by Legal Daily (reprinted by China Media Group / CNR), National Business Daily, Guangming Daily, and Xinhua-affiliated reports show that commercial "AI-lowering" services advertise claims like "pure manual" (纯人工) and "guaranteed pass" (包过)—with one reported seller promising "sole individual operation, management science PhD graduated" (纯个人操作，管科博士已毕业)—yet in practice rely on automated prompt templates or mechanical phrase manipulation like synonym substitution, word-order inversion, and splitting long sentences. This frequently damages academic rigor, introduces logical contradictions, and in documented cases causes AI scores to rise rather than fall.

Beyond poor editorial quality, relying on unregulated commercial intermediaries creates serious vulnerabilities: extortion where operators threaten to send reporting letters to universities or leak manuscripts online, severe personal data exposure, and potential violations of institutional guidelines regarding outside assistance and authorship integrity.

Documented cases indicate that turning to commercial AI-lowering intermediaries rarely resolves screening concerns. Instead, students risk compounded problems: distorted prose that fails faculty review, contractual extortion, and administrative complications arising from compromised drafting records.

Evaluating commercial thesis modification services requires examining documented public investigations rather than promotional marketing copy. Multiple investigative reports published between 2024 and 2026 describe the structural mechanics and consumer harms of this industry:

1. Legal Daily (reprinted by CNR): Misrepresented manual work, fee disputes, and explicit extortion

An extensive investigation by Legal Daily (reprinted by China Media Group / CNR) titled Investigation into the Chaos of Thesis AI Rate Lowering Services: Students Face Fraud and Even Threats (代降论文AI率服务乱象调查：学生遭遇欺诈还被威胁, 2025-07-14) documented student experiences where fees ranged from tens to over a thousand yuan (such as a 100-yuan deposit, 150 yuan, 180 yuan, or over 500 yuan).

The investigation revealed that:

• Misrepresented manual work: In one documented case, a vendor promised "sole individual operation, management science PhD graduated" (纯个人操作，管科博士已毕业); other reported cases revealed operators providing instructions on using AI prompts or running text through automated prompts. In documented student experiences, AI detection metrics rose rather than fell (in one case increasing from 35% to 88%).
• Explicit extortion threats: When students requested refunds over unusable drafts or failed pre-checks, rogue operators refused refunds and issued direct threats. The report recorded two explicit extortion demands: "If you don't pay the balance, I will upload your thesis across all platforms" (不付尾款我就帮你全平台上传论文), and threatening that "an electronic report letter will be sent immediately to your university mailbox and the Ministry of Education" (电子版举报信马上发到学校邮箱和教育部).

2. National Business Daily & Xinhua: Grey market "lowering-writing" and ghostwriting realities

An undercover investigation by National Business Daily (高校严查论文AI率，帮学生“降写”成了火爆生意, 2026-06-10) and coverage distributed by Xinhua (论文AIGC率，检测标准科学吗？, 2026-05-23) examined the commercial mechanics of these intermediary services:

• Deceptive "guaranteed pass" marketing: Undercover reporting by National Business Daily revealed that a vendor (selling a rewriting system with 4,166 sales) claimed: "guaranteed reduction; CNKI, VIP, and Gezida all pass in one attempt; 90% of users end up around 10% after reduction" (保证降，知网、维普、格子达都是一次过，90%的用户降完后都是10%左右).
• Concealing academic ghostwriting: Legal counsel cited in the National Business Daily report emphasized that such operations operate "in the name of technical services to conceal the reality of ghostwriting" (以技术服务之名，掩盖代写之实).

3. Guangming Daily: Mechanical rewriting and compounding risks

An investigative report published in Guangming Daily (用AI降AI率，降出了什么, 2026-05-28, p. 08) analyzed the actual methods deployed by commercial services and their consequences:

• Mechanical phrase manipulation: While vendors advertise "pure manual" (纯人工), "guaranteed pass" (包过), and "worry-free after-sales" (售后无忧), the report observed that intermediary modifications rely primarily on "synonym substitution, word-order inversion, and splitting long sentences" (近义词替换、语序倒装、长句拆分).
• Escalating student risks: Rather than resolving screening concerns, multiple teachers cited in the report warned students of risks including "AI rates rising instead of falling, manuscript leakage, and extortion" (AI率不降反升、论文泄露、被勒索胁迫). Attorney Wang Weiwei cited in the report stated that purchasing third-party services like ghostwriting or rewriting to lower AI rates "will constitute clear academic misconduct" (都会构成明确的学术不端行为), and cautioned that in serious circumstances such commercial operations "may constitute illegal business operation" (严重的可能构成非法经营罪).

Note on objective evaluation: These published reports characterize prevailing commercial practices across the intermediary sector. Without verifiable primary evidence regarding an individual business entity, users should assess services based on structural contract transparency and operational compliance rather than unsubstantiated claims.

When assessing any external service, platform, or independent contractor offering editorial assistance, the presence of any of the following five practices indicates unacceptable academic, legal, or financial risk:

1. Absolute score guarantees and "guaranteed pass" promises

Promising that a manuscript will "reach below 10%" or claiming, as a vendor in the National Business Daily investigation stated, that "CNKI, VIP, and Gezida all pass in one attempt," is mechanically impossible and fundamentally misleading:

• AIGC detection tools rely on probabilistic predictive models that vary significantly across platforms (such as CNKI, VIP, Wanfang, or Turnitin).
• Model parameters and database baselines update continuously; a text showing 8% on an unverified third-party platform may produce an entirely different figure on an institution's official screening portal.
• Legitimate academic editors provide editorial clarity and linguistic refinement; they do not guarantee arbitrary mechanical detection scores.

2. Demands for institutional credentials or library intranet logins

Any vendor requesting your university portal username, campus intranet password, or institutional library proxy to "perform an official check" must be rejected immediately:

• Handing over campus credentials violates institutional computer use and cybersecurity policies at many universities.
• It exposes your full academic history, course records, and personal identity to unknown third parties.
• Sharing credentials creates severe security risks, including the potential unauthorized exploitation of institutional checking allowances or the theft of unpublished manuscripts.

3. Ambiguous data retention, storage, and privacy terms

Submitting a complete thesis draft without a binding, transparent data-handling policy creates permanent vulnerability:

• Unregulated vendors may retain, aggregate, or resell your research data, preliminary findings, or complete manuscript to paper mills or third-party training databases.
• If an intermediary stores your text in an unverified repository or database, subsequent institutional submissions may trigger severe similarity flags for self-collusion or unoriginal content. For an analysis of how third-party checking platforms handle repository storage and data retention, see our guide on whether third-party Turnitin checks store your paper.
• As demonstrated in published investigations, holding an unreleased manuscript gives predatory operators leverage for post-transaction extortion.

4. Demands for full payment prior to an agreed, written scope of work

Transactions conducted via informal chat channels without a detailed, legally binding scope of work represent extreme consumer hazards:

• Vague terms such as "comprehensive AI reduction" allow vendors to claim completion after running a single automated script.
• Predatory operators frequently demand arbitrary additional surcharges before releasing the revised file.
• Without a written service agreement defining line-by-line editorial boundaries, students have no viable recourse through consumer protection bodies or legal channels.

5. Advising students to falsify, tamper with, or delete writing history

Any service suggesting that you delete version history, clear cloud document edit logs, or fabricate timestamped notes is advising you to destroy your defense:

• When a university committee reviews an elevated AI score, the primary exonerating evidence is the authentic, continuous history of creation (incremental drafts, timestamped research logs, reference notes, and edit records).
• Deleting your actual drafting timeline or attempting to simulate artificial edit histories creates affirmative evidence of deliberate deception, escalating a technical screening question into serious administrative suspicion when evaluating original authorship.

Beyond financial risks, students must understand how academic institutions evaluate original work and external editorial support. Policies regarding writing tools, polishing, and third-party services vary across universities, and navigating them requires understanding three core principles:

Legitimate academic work relies on genuine authorship and verifiable drafting records. Before using any external editing, rewriting, or polishing tool, students should consult their institution's specific written guidelines rather than accepting an intermediary's commercial assurances.

1. Screening thresholds versus misconduct determinations

Institutional AIGC metrics (such as a 15%, 20%, or 40% screening line) are initial administrative triage indicators, not definitive proof of academic misconduct. Official notices from universities (such as Tianjin University of Science and Technology and Aba Teachers University) demonstrate that institutions establish specific appeal and review procedures, allowing students to demonstrate genuine authorship through supervisor verification or drafting materials if an initial score is elevated.

2. Permissible assistance depends on institutional and assignment rules

Commercial services often market their offerings under ambiguous labels such as "manual polishing" or "safe rewriting." However, what forms of editing, rewriting, or automated assistance are permitted is governed strictly by the written policies of the student's university, department, and specific degree committee. Under Chinese degree regulations, including Article 37 of the Academic Degrees Law (中华人民共和国学位法), acts such as "ghostwriting, plagiarism, and forgery" (代写、剽窃、伪造等) are strictly prohibited violations of academic integrity. For editorial polishing, translation, and AI assistance tools, individual institutions establish distinct standards. Some programs permit copyediting and require formal disclosure in an acknowledgment section, while others restrict external interventions more narrowly. Readers should never rely on an intermediary's claims; what is allowed depends entirely on their institution's and assignment's written rules.

3. The decisive role of authentic process evidence

During an academic inquiry, evaluation committees do not rely on commercial third-party "clearance certificates." Instead, committees review:

• The continuity of rough drafts, outline iterations, and supervisor consultation notes;
• Document version histories displaying progressive, human writing habits over weeks or months;
• The student's ability to defend the methodology, citations, and specific wording orally.

Hiring an intermediary to tamper with the text severs the author's organic connection to their own manuscript, leaving them unable to explain aberrant syntax or terminology during an oral inquiry.

If you are considering any professional editorial, translation, or linguistic support for your thesis, complete this due diligence review across four essential categories before sharing any text:

1. School rules and departmental policies

• [ ] Review the current-year degree handbook: Verify whether your specific department or college permits third-party language polishing, rewriting assistance, or editorial services, and whether formal acknowledgment is required.
• [ ] Check supervisor requirements: Clarify with your faculty advisor what forms of editorial or technical assistance are considered appropriate within your academic discipline.
• [ ] Identify the official screening system: Confirm which detection platform (e.g., CNKI, VIP, Wanfang) your university officially designates, avoiding reliance on external proxy scores.

2. Authorship integrity and process documentation

• [ ] Preserve all incremental versions: Save dated milestone drafts, research notes, and initial outlines in independent, secure personal cloud storage.
• [ ] Retain native platform version histories: Keep uninterrupted edit histories within your word processor (e.g., Word Version History, WPS History, or Google Docs Version History).
• [ ] Retain full editorial agency: Never permit an external party to modify substantive arguments, findings, or interpretations; ensure every final wording choice remains your own.

3. Data privacy and manuscript security

• [ ] Examine data retention agreements: Ensure the provider explicitly guarantees in writing that manuscripts are not permanently stored, indexed, shared, or incorporated into commercial training sets.
• [ ] Anonymize identifying information: Strip institutional affiliations, student ID numbers, mentor names, and grant identifiers prior to transmitting files for technical proofreading.
• [ ] Never disclose institutional credentials: Maintain complete separation between external editing transactions and your university network accounts.

4. Contractual clarity and scope definition

• [ ] Establish written scope in advance: Insist on a formal contract or written statement of work detailing exactly what services are provided (e.g., grammar correction vs. structural feedback).
• [ ] Reject open-ended or arbitrary pricing: Avoid services operating on arbitrary, post-submission fee adjustments or unitemized surcharges.
• [ ] Reject unscientific guarantees: Decline any service that promises specific numerical detection scores, "undetectable" outputs, or guaranteed passing verdicts.

Does paying for "manual AI lowering" guarantee passing an institutional AI review? No. Published investigations by Legal Daily, National Business Daily, and Guangming Daily confirm that commercial services claiming to guarantee specific scores are marketing fabrications. Detection models are probabilistic, vary across platforms, and undergo continual algorithm adjustments. Furthermore, mechanical phrase manipulation (such as synonym substitution or sentence inversion) by non-specialist intermediaries frequently introduces syntactic distortions that supervisors reject during regular thesis reviews.

What should I do if an intermediary threatens to report my thesis after a payment dispute? Do not submit to extortion or make additional panic payments. Retain all communication logs, payment receipts, service descriptions, and original, timestamped manuscript files. In published reporting, legal counsel noted that such intermediary operations operate in the name of technical services to conceal ghostwriting, and in serious circumstances may potentially constitute illegal business operations. Because rogue operators exploit student anxiety rather than lawful standing, preserve your drafting records, refrain from paying additional demands, and seek guidance from university student legal aid or counseling resources to demonstrate authentic authorship.

Can an intermediary provide an "official" CNKI or VIP AI report before university submission? It is generally unverified and inadvisable. Commercial resellers operating on e-commerce platforms cannot reliably verify authorized institutional administrative access. Commercial reports obtained through third-party proxies frequently run on different database baselines, uncalibrated consumer editions, or alternative algorithms that do not match the institutional server settings used during formal degree clearance.

How should I address an elevated AI score without hiring an intermediary? Begin by reviewing your department's formal policy. An elevated initial score is a prompt for review, not an automatic misconduct conviction. Prepare your primary evidence: incremental working drafts, lab notebooks, supervisor email threads, and version history logs. Schedule a direct discussion with your faculty advisor, explain your actual drafting methodology, and review any flagged paragraphs together to confirm whether standard disciplinary phrasing or template formats triggered a probabilistic anomaly.

The analysis in this guide is drawn from published investigative journalism, academic integrity policies, and consumer protection records:

1. Legal Daily (法治日报, reprinted by CNR / 央广网): 代降论文AI率服务乱象调查：学生遭遇欺诈还被威胁 (Investigation into the Chaos of Thesis AI Rate Lowering Services: Students Face Fraud and Even Threats, 2025-07-14) URL: `https://news.cnr.cn/native/gd/20250714/t20250714_527260647.shtml`
2. National Business Daily (每日经济新闻): 高校严查论文AI率，帮学生“降写”成了火爆生意 (Universities Strictly Check Thesis AI Rates: Helping Students 'Lower and Ghostwrite' Becomes Booming Business, 2026-06-10) URL: `https://www.nbd.com.cn/articles/2026-06-10/4422770.html` (Reprinted by Sina: `https://finance.sina.com.cn/roll/2026-06-10/doc-iniaxvfk2625001.shtml`)
3. Xinhua News Agency (新华社 WeChat account, via CCTV News / Quanzhou Web): 论文AIGC率，检测标准科学吗？ (Thesis AIGC Rate: Is the Detection Standard Scientific?, 2026-05-23) URL: `https://www.qzwb.com/gb/content/2026-05/23/content_9232318.htm`
4. Guangming Daily (光明日报): 用AI降AI率，降出了什么 (Using AI to Lower AI Rates: What Does It Produce?, 2026-05-28, p. 08) URL: `https://epaper.gmw.cn/gmrb/html/content/202605/28/content_15178.html` (Reprinted by Sina / Elephant News: `https://finance.sina.com.cn/stock/t/2026-05-28/doc-inhzmpqq7861953.shtml`)
5. Beijing Daily (北京日报): 论文AI检测，正在“逼疯”这届大学生 (Thesis AI Detection Is Pushing This Year's Graduates to the Brink, 2026-07-19 21:05) URL: `https://xinwen.bjd.com.cn/content/s6a5ccb7de4b03fa51a81b89b.html`
6. Standing Committee of the National People's Congress of the PRC: 中华人民共和国学位法 (Academic Degrees Law of the People's Republic of China, Article 37 provisions on ghostwriting, plagiarism, and forgery, 2024-04-26) URL: `http://www.npc.gov.cn/npc/c2/c30834/202404/t20240426_436840.html`
7. Institutional Review & Appeal Guidelines: Notices on thesis AIGC review and appeal procedures from Tianjin University of Science and Technology (`https://jw.tust.edu.cn/ggtz/65b256ce743a41fa82c3533e16009807.html`) and Aba Teachers University (`https://jw.abtu.edu.cn/info/1131/4590.htm`).