AI Usage Policy

PraxiHub permits responsible use of artificial intelligence in selected research, analysis, editorial, technical, and publication workflows. AI may assist the process. Humans remain responsible for the research.

This policy explains the standards that apply when AI tools are used in connection with PraxiHub Research, Review Articles, Research Summaries, Datasets, Knowledge resources, figures, code, and other publication materials.

The purpose of this policy is to support transparent and responsible use of AI without allowing automation to replace human authorship, source verification, methodological judgment, interpretation, or accountability. AI assistance does not change PraxiHub's requirements for evidence, citation integrity, reproducibility, originality, privacy, disclosure, and accurate content classification.

• Human accountability: A named human author or responsible contributor must review and accept responsibility for published work.
• Source verification: AI-generated statements and references must be checked against original or authoritative sources.
• Transparency: Material generative-AI assistance must be disclosed accurately.
• No fabricated evidence: AI must not be used to invent references, data, findings, quotations, observations, or methods.
• No AI authorship: An AI system cannot be an Author, Co-author, Researcher, or scholarly contributor.
• Human methodological judgment: AI output cannot substitute for validation of research design, calculations, code, analysis, or interpretation.
• Privacy and confidentiality: Private, confidential, proprietary, sensitive, or embargoed material must be protected.
• Proportionate review: The level of human verification must reflect the risk and importance of the AI-assisted task.

Subject to human review, AI may assist with:

• grammar correction, language polishing, clarity, tone, translation, and prose restructuring;
• brainstorming, outlining, and summarizing material supplied and verified by a researcher;
• literature-discovery and citation-format assistance;
• documentation, formatting, and schema preparation;
• code assistance, debugging, and validation-script development;
• data-cleaning workflow design; and
• chart or visualization code.

Permission to use AI does not mean that its output may be accepted without verification. Researchers and editors must assess whether the output is accurate, appropriate, original, secure, and consistent with the underlying evidence.

PraxiHub prohibits using AI to create or present:

• fabricated references or invented DOIs;
• invented quotations, URLs, authors, affiliations, or credentials;
• fabricated datasets, observations, interview responses, or participant data;
• fake statistical results or unsupported empirical findings;
• synthetic evidence presented as authentic evidence;
• unsupported claims that empirical testing occurred;
• fabricated systematic-review, peer-review, or research-methodology claims;
• copied or protected material disguised through automated paraphrasing; or
• AI-generated people presented as real researchers or contributors.

AI must not be used to conceal plagiarism, evade attribution, manufacture consensus, or misrepresent the origin or reliability of content.

For Original Research, AI may assist with code, language, formatting, documentation, or exploratory analytical work. Human researchers must independently verify research design and methodology, equations and assumptions, statistical procedures, data transformations, calculations and code execution, reported results, interpretation, and conclusions.

AI-generated empirical results must not be published without reproducible human verification.

For Review Articles and Conceptual Reviews, AI may assist with organization, language, source discovery, or synthesis structure. Authors must verify every source and ensure that claims accurately reflect the cited literature. AI must not invent scholarly consensus or a review methodology.

A publication must not be described as a systematic review, empirical study, or peer-reviewed work unless the documented process supports that classification.

Research Summaries must remain faithful to the original work. AI-assisted summarization is permitted only when the original source has been reviewed, the interpretation has been checked, attribution is clear, and no new finding is falsely attributed to the source.

For Research Datasets, AI may assist with code generation, data-cleaning workflows, schema preparation, documentation, and validation or reproducibility scripts.

AI must not silently fabricate records, replace missing empirical observations, interpolate missing values without disclosure, alter raw observations without a documented method, create false provenance, or present synthetic records as observed data.

Any AI-assisted transformation that materially affects published data must be documented, reproducible, and reviewed by a responsible human. Dataset provenance must come from verifiable data sources and processing pipelines. AI-generated output cannot substitute for actual provenance.

When AI assists with analytical code, the researcher must validate the equations, assumptions, implementation, execution, outputs, statistical meaning, security, licensing, and reproducibility. Successful code execution alone is not sufficient evidence that an analysis is correct.

AI-generated output must not be treated as independent evidence for a substantive scholarly claim unless the AI system or its output is itself the object of study and the methodology explicitly documents that use. Otherwise, claims must be supported by original or authoritative sources.

References, quotations, statistics, citations, URLs, DOIs, datasets, and factual assertions generated or suggested by AI must be verified against original or authoritative sources. The underlying source—not the AI system—must be cited as evidence.

Researchers must confirm that a cited source supports the actual proposition, context, date, unit, and strength of wording used in the publication. Correlation must not be presented as causation, and a source-specific finding must not be generalized beyond its evidence.

Where AI materially assists a research workflow, that assistance should be disclosed under this policy rather than substituted for scholarly citation.

AI-assisted writing and editing are permitted for tasks such as grammar, clarity, restructuring, translation, and tone.

The human author must review and approve the final text and remains responsible for every published sentence. AI assistance does not excuse inaccurate language, unsupported certainty, plagiarism, improper attribution, or misleading interpretation.

Material supplied by an AI system must be checked for factual errors, hidden assumptions, incomplete context, invented sources, and inappropriate similarity to third-party work.

AI systems cannot be listed as an Author, Co-author, Researcher, or contributor carrying scholarly responsibility. Authorship requires human responsibility and accountability.

Every named author remains responsible for accuracy, evidence and sources, methodology, data and calculations, interpretation, citations and attribution, conclusions, permissions and originality, ethical and legal compliance, disclosure of limitations and material AI assistance, and the final published language.

Use of AI does not transfer or reduce this responsibility.

Disclosure is not normally required for trivial automated assistance such as ordinary spellcheck, basic grammar correction, formatting, or standard autocomplete.

Disclosure is required when generative AI materially contributes to drafting or restructuring substantive sections, literature synthesis, code creation, analytical workflows, interpretation support, figure or visualization generation, or research-design assistance.

PraxiHub Research publications may use an AI-Assisted Research Process section or another appropriately labeled disclosure. The disclosure must reflect the actual workflow and must not be copied into a publication when it would be false or misleading.

A suitable disclosure may explain which stages received AI assistance and must make clear that human authors verified the sources, claims, methods, interpretation, and final content.

Existing publications must not receive retroactive AI disclosures unless their actual workflow has been reviewed and supports the disclosure.

Future contributor or submission workflows may require acknowledgment of this policy, including: I have read and agree to PraxiHub's AI Usage Policy. This statement does not imply that a public submission or peer-review system currently exists.

AI-generated or AI-edited images must not be presented as factual research evidence unless their use is scientifically appropriate and clearly disclosed.

Synthetic material must not masquerade as an empirical observation, original photograph, experimental evidence, or authentic researcher portrait.

Researcher and Team portraits must represent real individuals. AI must not replace or materially alter identity in a misleading way. Technical processing such as cropping, resizing, compression, metadata removal, and limited color correction is permitted only when it does not materially alter a person's identity, appearance, or evidentiary meaning.

AI may assist with charts, figures, or visualization code only when the underlying data are authentic, formulas and transformations are verified, labels and scales are accurate, the source and methodology remain transparent, and the visualization does not imply unsupported evidence.

Researchers must not submit confidential, proprietary, embargoed, private, or personally sensitive information to an external AI system unless they have authorization and the system's data handling is appropriate for the material.

Protected information includes private email addresses, phone numbers, physical addresses, passwords, tokens, credentials, private keys, confidential participant information, unpublished sensitive datasets, proprietary code or documentation, embargoed manuscripts, and confidential reviewer comments.

Particular care is required for unpublished manuscripts, proprietary datasets, confidential research, and reviewer material.

If PraxiHub introduces peer review in the future, confidential manuscripts and reviewer materials must not automatically be uploaded to public AI systems. This policy does not claim that PraxiHub currently operates an external peer-review system.

Researchers and contributors using AI must understand the task for which AI is used, review and verify the resulting output, preserve evidence and source provenance, document material analytical transformations, test critical code independently, comply with relevant confidentiality and licensing requirements, disclose material generative-AI assistance, and correct errors discovered after publication.

AI use does not excuse plagiarism. Generative systems must not be used to disguise copied work, evade attribution, or reproduce protected content improperly.

PraxiHub does not treat AI-detection scores as conclusive evidence of authorship or misconduct. Publication decisions should focus on evidence, methodology, citation integrity, originality, transparency, and documented process.

PraxiHub may, where appropriate, request clarification about AI use, source verification, relevant code, calculations, or analytical documentation; require an accurate disclosure or correction; or reject or withdraw material that does not meet research-integrity requirements.

If undisclosed or inappropriate AI use materially affects accuracy or integrity, PraxiHub may correct the publication, add a reviewed disclosure or editorial note, or withdraw the publication where necessary.

Enforcement should be proportionate to the issue and should preserve an accurate publication history. Questions or concerns about compliance may be submitted through PraxiHub's established Corrections and Contact channels.

This policy is owned by PraxiHub and may be updated as research practices, publication workflows, and AI capabilities evolve.

Material updates will be reflected through a revised version number and effective date. An update to this policy does not by itself rewrite historical publication disclosures or change the documented workflow of an existing publication.