JAIGRJournal of AI-Generated Reviews

Editorial policy

Last updated 2 November 2026

Scope

Journal of AI-Generated Reviews publishes literature reviews whose scholarly content was generated by AI systems, in all subject areas, provided they meet the content standards below. Eligibility is decided reference by reference: every cited work must have a DOI and an open-access full text that we can retrieve legally.

AI-generated reviews only

The journal accepts only literature reviews whose scholarly content was generated by an AI system. Papers written by humans, including human-written papers lightly edited by AI, are rejected. A false declaration of AI authorship discovered after publication leads to retraction.

Verified authors

Every listed author signs in, verifies an ORCID iD and a university email address (from a degree-granting institution), and confirms their authorship. Author names and affiliations are taken from the authors' ORCID records. A submission's checks start only when every author has confirmed; if any invited author declines, the submission is withdrawn.

What makes a submission eligible

  • The scholarly content was generated by AI. The submitting person declares the model, version, generation date and the prompts or workflow used, and declares that they did not write the content themselves.
  • Every claim that relies on the literature cites it with numbered citations such as [3] or [2, 5–7], and every listed reference is cited.
  • Every reference has a DOI registered with Crossref or DataCite, the cited title matches the registered title, and an open-access full text is available.

Human accountability

AI systems cannot be accountable for a publication. Two people are: the submitting author, who answers for the declarations they make and is named on the article; and the responsible editor, Dr. Ding Han, who decides what is published, handles appeals, and issues corrections and retractions.

What the assessment is, and is not

Articles are AI-assessed, not peer-reviewed. Each citation-bearing sentence is checked automatically against the full text of the paper it cites, and the result is published with the article. The assessment checks fidelity to the cited sources only. It does not assess novelty, completeness of coverage, methodology or importance, and it can be wrong. Readers should treat the verdicts as evidence to inspect, not as a seal of quality. How the assessment works.

Content standards

The journal publishes only uncontroversial scholarly reviews. We do not publish content that may breach applicable law, content on controversial or divisive subjects, violent, sexual, hateful, harassing or defamatory content, or misinformation. Every submission is screened at submission and again during the weekly verification, and the editor cannot publish a submission unless both screens are clear. When in doubt, we do not publish.

Desk rejection

A submission is desk-rejected, without review, when:

  • the reference list has errors: a DOI that is not registered, a title that does not match the registered work, a cited work without an open-access full text, a citation to a reference that is not listed, or a listed reference that is never cited;
  • it is of low quality: shorter than 1200 words, citing fewer than 8 works or making fewer than 10 cited claims, or judged below the journal's minimum scholarly quality (scope, synthesis, coverage, accuracy, structure, language and originality) by the AI quality review;
  • it shows plagiarism: more than 10% of its unquoted text, or any unquoted passage of more than 30 words, copied verbatim from the cited sources or from another submission to this journal;
  • it appears to be written by a human: the journal accepts only AI-generated reviews. The weekly review assesses authorship; because no automated test of authorship is certain, the editor reconsiders on appeal;
  • the content-safety review recommends rejection, or the editor finds it outside the journal's scope or standards.

Review rounds

Each submission receives at most two rounds of review. Each round consists of the automated checks, the weekly AI verification of every cited claim, the content and quality reviews, and a decision by the editor.

  1. Round 1: minor revision, major revision, or reject. The decision letter explains what must change.
  2. Round 2 (after the author's revision, with a response to the decision letter): accept or reject. There is no further revision.

A submission cannot be accepted while any cited claim is assessed as contradicted or not supported by its source, or while either content screen or the quality review is not clear.

Resubmission after rejection

A rejected or desk-rejected paper may be resubmitted once, as a new submission, from seven days after the decision. Submitting substantially the same text earlier is refused automatically.

Process and timing

  1. Reference, plagiarism, length and topic checks run within minutes of submission.
  2. Submissions that pass are verified in a weekly batch. If the weekly capacity is full, a submission waits for the next week.
  3. The editor decides as described above. Accepted articles are published with a permanent article number and URL.

Fees and access

There are no submission or publication fees. All articles are free to read.

Versions of record

When an article is published, its text, references and assessment are frozen as version 1, with a SHA-256 checksum shown on the article page. Later corrections to metadata create new versions; earlier versions remain readable. Errors in the review text are documented by an editorial notice, or a substantially revised review is assessed as a new submission. See corrections and retractions.

Indexing and identifiers

Articles are identified by journal article numbers (for example JAIGR.2026.0001). The journal does not yet assign DOIs and does not claim inclusion in any index.