Publication policy
Editorial standards
Effective 18 September 2026. AI Signal separates verified facts, interpretation, and conditional outlook. Demonstration material is prominently labeled and must never be presented as reporting.
Sourcing and verification
- Material factual claims require identifiable sources and verification dates.
- Primary sources are preferred. A source’s assertion is attributed and is not treated as independent proof.
- Anonymous sourcing is exceptional and requires the responsible editor to document the public-interest case and corroboration privately.
- Quotes, images, datasets, and documents must have a lawful basis for publication and accurate attribution.
Review and fairness
Reported stories require recorded human editorial approval. Serious criticism should be specific, evidenced, proportionate, and normally give the affected party a reasonable opportunity to respond before publication. Headlines and signal labels must reflect the article’s evidence.
Analysis, uncertainty, and conflicts
Analysis is labeled as judgment. Predictions remain conditional. Material financial, professional, personal, or commercial conflicts connected to a story must be disclosed. Sponsored influence over editorial conclusions is not accepted.
AI assistance
AI tools may assist research organization, drafting, coding, and editing. They are not treated as sources. A human editor remains responsible for verification, originality, rights, fairness, and the publication decision. Sensitive or confidential source material must not be entered into an AI service without authority and appropriate safeguards.
Corrections and contact
Substantive errors follow the corrections policy. Raise an editorial concern through the repository’s public issue tracker; use the private security route for vulnerabilities.
This policy describes editorial practice. It does not promise coverage, publication, or a particular conclusion.