Artificial Intelligence (AI) Policy & Operational Guidelines

• Reporters At Large (reportersatlarge.com)
• Version 1.1
• Effective: December 1st, 2026
• Next review: June 1st, 2027

Summary

Reporters At Large uses AI tools only to assist human journalists, never to replace them. A named editor is accountable for every published story. Anything AI-generated or heavily AI-processed is labelled. Errors are corrected openly, and readers can complain to a named contact.

1. Purpose and Scope

This policy sets the rules for using AI in how we gather, write, edit, produce and distribute news. It protects accuracy, independence, source confidentiality and reader trust.

It binds all full-time, part-time, contract and freelance staff, including reporters, columnists, editors, producers, photographers, designers, researchers, technical staff and interns. It applies to every kind of content we publish:

• News reports, features, analysis and explainers
• Opinion pieces and guest contributions
• Press releases and sponsored or promotional content
• Photos, graphics, video, audio and RAL TV output
• Headlines, summaries, social posts and newsletters

2. Key Definitions

These terms are used consistently throughout this policy.

• AI-assisted: a human wrote or produced the content and used AI for support tasks such as spell-checking, transcription or research.
• AI-generated: substantial text, images, audio or video were produced by an AI system rather than by a person.
• Core facts: names, dates, figures, quotes, allegations, and any claim that could harm a person’s reputation or that the story depends on.
• Approved tools: the AI tools listed by the Policy Owner (see Section 9). Staff may not use unlisted tools for newsroom work.
• Sensitive information: confidential or off-the-record material, source identities, unpublished documents, and personal data.

3. Permitted Uses

AI may be used for the following, always under human review:

• Grammar, spelling and style checks
• Transcription, captions and format conversion
• Summarising public documents and reports to speed up human reading
• Brainstorming angles and story ideas
• Translation, with review by a fluent staff member or editor
• Analysing public datasets to flag leads for human follow-up
• Comment moderation and SEO metadata suggestions
• Background research, where every fact is then checked against a primary source

4. Prohibited Uses

The following are not allowed under any circumstances:

• Publishing AI-generated news articles, features, opinion or analysis as our own reporting, whether or not an editor skims them.
• Connecting any AI system to the CMS so that it can publish without a human decision.
• Using AI to confirm core facts, verify anonymous sources or reach the conclusions of an investigation. AI may point to leads, but a human must verify them against primary sources.
• Creating photorealistic images, video or audio of real events, places or people, including deepfakes and synthetic voices that imitate a real person.
• Creating fake sources, quotes, bylines or author personas.
• Entering sensitive information into any AI tool that is not approved for it (see Section 8).

5. Human Oversight and Accountability

• Every story has a named accountable editor who approves it before publication.
• The reporter and the approving editor are jointly responsible for accuracy, fairness and legal compliance. AI cannot be blamed for an error.
• Any fact, quote, statistic or citation suggested by AI must be independently verified with a primary source before use. Publishing a fabricated AI fact is treated as editorial negligence and handled under Section 10.
• Staff who are unsure whether a use is permitted must ask the Policy Owner before proceeding.

6. Transparency and Labelling

Routine assistance does not need a public label, so spell-checking, transcription and grammar tools are exempt. The following must be labelled in a visible note on the page:

• AI-generated or AI-edited images and graphics: “This image was created or altered with AI assistance and reviewed by Reporters At Large editors.”
• Machine-translated articles or sections: “This article was translated with AI assistance and reviewed by our editors.”
• Data analysis or visualisations where AI materially shaped the output: “AI tools were used to process the underlying data. Findings were verified and edited by our editorial team.”
• Synthetic voiceovers or AI-generated audio or video elements: a clear on-screen or in-text disclosure.

Where our own translation has been reviewed by a fluent person, the label is still required if AI produced the first draft of the published text.

Sponsored and promotional content must be labelled as such, whether or not AI was involved. AI must not be used to disguise paid or partisan material as independent reporting.

7. Fairness, Bias and Local Context

AI systems can reflect biases in their training data, including Western-centric and linguistic biases. When using AI on stories about Nigerian and African communities, politics or languages, editors must check the output for stereotypes, mistranslations and missing local context. We review how AI is used in our coverage at each policy review and welcome reader feedback about bias (see Section 10).

8. Privacy, Sources and Data

• Never enter sensitive information into public or consumer AI tools such as ChatGPT, Claude or Gemini unless the tool is approved for that purpose and the Policy Owner has confirmed its data handling terms.
• Only public documents and published material may be used with general-purpose AI tools.
• Source identities and unpublished investigative material must stay out of AI tools entirely, unless the tool runs in a secured environment that does not retain or share the data. We will only claim such an environment if we have one.
• We handle personal data in line with the Nigeria Data Protection Act 2023 and any other law that applies to a given story.
• We follow the professional codes that apply to our work, including the Nigeria Union of Journalists (NUJ) code of ethics, and we are informed by international guidance such as the Paris Charter on AI and Journalism.

9. Policy Owner and Governance

• Policy Owner: Reporters At Large Editorial Team, Contact: news@reportersatlarge.com.
• The Policy Owner maintains the list of approved AI tools, answers staff questions, decides edge cases, and keeps a simple log of significant AI use and any incidents.
• Review: The policy is reviewed at least every six months, and sooner after a serious incident, a legal change or a major shift in AI technology. Each review updates the version table below.
• Training: staff receive an onboarding briefing on this policy and a refresher after each substantive update.
• Breaches are handled under Section 10 and may lead to correction or retraction of the content, retraining, suspension of AI tool access, or disciplinary action, depending on severity.

10. Complaints and Corrections

How to raise a concern

Anyone, whether a reader, a person named in a story, or a staff member, can report an AI-related concern, a factual error, a missing label or suspected AI misuse.

• Email: [ai-policy@reportersatlarge.com]
• Contact form: Contact
• Please include the page link, a description of the problem and, if possible, supporting evidence.

What happens next?

1. Acknowledgement: we confirm receipt within [2] working days.
2. Review: the Policy Owner, or the editor responsible for the story, assesses the complaint within [5] working days. Complaints involving serious harm, such as defamation, safety risks or fabricated content, are escalated immediately.
3. Action: where an error is confirmed, we fix it promptly using the standards below.
4. Reply: we tell the complainant what we decided and why.

Correction standards

• Minor errors (spelling, a wrong caption, a small factual slip): corrected in the text, with a note at the bottom of the article stating what changed and when.
• Material errors (a wrong fact, name or figure that affects meaning): corrected in the text, with a visible correction notice at the top or bottom of the article. If the story was shared on our social channels, a correction is posted there too.
• Serious errors or fabricated AI content: the content may be removed or retracted, with a public notice explaining why. An internal review follows to prevent a repeat.
• Corrections are never silently deleted, and original reporting is not quietly rewritten to hide a mistake.
• Every AI-related correction is recorded in the Policy Owner’s log and considered at the next review.