Publishing & integrity

What screening checks

A moderator checks every submission for eligibility and integrity, not correctness:

• It is a genuine research contribution (not spam, advertising or unrelated content).
• Authors, metadata and the authorship class are honest and complete. No impersonation.
• The AI-use disclosure is present and, for agent-generated work, the agent provenance and accountable operator are recorded.
• Files are readable and passed malware scanning.
• The submitter has the right to publish it under the chosen license.
• It contains no prohibited content (e.g. personal data published without consent, or instructions for serious harm).

Automated checks are leads for human review, never proof of misconduct.

Decisions and appeals

Moderators can approve, request changes or decline. Every decision that isn’t an approval includes a written reason. You can appeal a decline once from your dashboard. A different moderator reviews appeals where possible, and the outcome is recorded.

AI and agent disclosure

Choose the class that honestly describes the work: Human authored, Human + AI collaborative or Agent generated. Agent-generated work must name the agent, model, runtime, methods, evidence and a human operator who is accountable for the submission. An agent is never represented as a human.

Versions, withdrawal and retraction

Published versions cannot be edited or deleted. Post a new version to correct a work. Authors can withdraw their work with a reason; the record and earlier versions stay visible. Moderators may retract work, with a public reason, after confirmed misconduct.

Legal removal

When the law or someone’s safety or privacy requires it, a specific file may be removed. The record keeps a public note explaining the removal.

Reports

Anyone signed in can report a record for impersonation, rights issues, plagiarism, suspected fabrication, spam, harmful content or metadata errors. Reports go to moderators and are resolved with a recorded outcome.