Artificial Intelligence, Automation & Synthetic Media Policy

1. Introduction

Artificial intelligence and automated technologies are changing how information is researched, produced, distributed and consumed.

VALCORED News Network recognises that these technologies can assist with certain editorial and administrative functions.

They can also produce fabricated facts, inaccurate summaries, misleading images, invented quotations, synthetic voices and material that appears credible despite having no reliable evidentiary foundation.

This policy defines VNN’s approach to responsible AI use while protecting journalistic accuracy, editorial independence, intellectual property and public trust.

Our guiding principle is:

Technology may assist journalism, but it must not replace responsibility for the truthfulness of what we publish.

2. Purpose

This policy establishes standards for:

• AI-assisted writing and editing.
• Automated news processing.
• Research and source verification.
• AI-generated images and video.
• Synthetic audio and voice technologies.
• Disclosure of material AI involvement.
• Human editorial responsibility.
• Privacy and confidential information.
• Copyright and licensing.
• Automated publication systems.
• AI-related corrections and accountability.

3. Scope

This policy applies to all AI-enabled systems used in connection with VNN’s editorial operations.

It covers employees, editors, contributors, contractors and authorised service providers.

Relevant technologies may include:

• Generative language models.
• Automated text-generation tools.
• AI-assisted search and research.
• Transcription and translation systems.
• Image-generation and editing systems.
• Synthetic voice and video tools.
• Automated news feeds.
• Content classification systems.
• Headline-generation tools.
• AI summarisation systems.
• Workflow automation and editorial plugins.

4. Human Editorial Accountability

VNN remains responsible for material it chooses to publish.

An AI system cannot independently assume the professional obligations of a journalist, editor or publisher.

Where AI substantially contributes to editorial content, the publication should remain subject to appropriate human oversight.

Human oversight should include consideration of:

1. Material factual accuracy.
2. Evidence and sourcing.
3. Fairness and relevant context.
4. Legal and ethical concerns.
5. Copyright and usage rights.
6. Sensitive personal information.
7. Misleading synthetic elements.
8. Necessary reader disclosures.

No content should be described as human-reviewed or independently verified unless the claimed review actually occurred.

5. Permitted AI-Assisted Activities

Subject to appropriate safeguards, AI tools may assist with the following functions.

5.1 Language and Readability

Improving grammar, spelling, structure and clarity without materially changing factual meaning.

5.2 Research Organisation

Organising research notes, identifying questions for further verification and creating preliminary summaries of source documents.

5.3 Transcription

Converting recordings into text for human verification and editing.

5.4 Translation

Assisting translation where original meaning is checked appropriately, particularly for consequential quotations.

5.5 Article Planning

Supporting the development of outlines, interview questions, explanatory structures and editorial workflows.

5.6 Headline Suggestions

Generating possible headlines that are subsequently reviewed for accuracy, context and journalistic suitability.

5.7 Content Classification

Assisting with categorisation, tagging, topic identification and other administrative functions.

5.8 Accessibility

Supporting alt-text preparation, captions and accessible presentation, subject to appropriate verification.

5.9 Technical and Administrative Functions

Supporting internal organisation, publication scheduling, formatting and appropriate content-management tasks.

Permission to use AI for these purposes does not mean that generated outputs may be published without review.

6. Prohibited AI Practices

VNN prohibits knowingly using AI to:

• Fabricate interviews.
• Invent quotations attributed to real people.
• Create non-existent eyewitnesses.
• Manufacture evidence.
• Invent documents or academic references.
• Attribute reporting to fictional journalists.
• Misrepresent synthetic images as authentic photographs.
• Falsely imply attendance at an event.
• Create deceptive impersonations.
• Present unverified AI claims as established facts.
• Conceal material sponsorship or editorial conflicts.
• Produce deliberately misleading news coverage.
• Evade applicable copyright, privacy or other legal obligations.

Any exception involving obviously fictional or satirical material must be labelled sufficiently to avoid misleading readers.

7. AI-Assisted News Article Preparation

Where an AI system assists with preparing editorial text, the responsible editor or contributor should ensure that material claims can be supported by reliable evidence.

The editorial process should distinguish between:

Source Material: Authentic documents, interviews, records or other evidence.

AI Output: Text produced or transformed by an automated system.

Verified Editorial Content: Material assessed through an appropriate editorial process.

AI-generated text must not be treated as equivalent to verified evidence.

A polished writing style does not establish reliability.

8. Source Verification and AI Hallucinations

AI systems may produce plausible but non-existent information.

Potential failures include:

• Incorrect names.
• Inaccurate dates.
• Fabricated official statements.
• Non-existent court decisions.
• Invented research citations.
• Misreported statistics.
• Misattributed quotations.
• Incorrect summaries of source documents.

Material factual assertions must be checked against suitable sources.

AI-generated references should be verified before being relied upon.

For high-impact stories, verification requirements should be stronger.

9. AI and Breaking News

Breaking news involves heightened uncertainty and potential consequences.

AI may assist with organising verified updates, but automated generation must not manufacture developments or fill information gaps with speculation.

Automated publication systems should not be granted unrestricted authority to present unverified information as confirmed.

Where AI is used to assist with breaking coverage, editorial controls should address:

1. The reliability of incoming sources.
2. The distinction between verified and preliminary information.
3. The risk of repeating outdated claims.
4. Proper attribution.
5. Human review requirements.
6. Escalation of high-risk material.
7. Correction and withdrawal procedures.

A developing event should not be described as resolved merely because an automated system produces a confident summary.

10. AI-Generated Images

Synthetic images may be appropriate for clearly identified illustrations, concept artwork or explanatory purposes.

However, they must not be presented as authentic documentary photographs of actual events.

Where an AI-generated image depicts an identifiable real person, location or event, VNN should assess the risk of misleading readers or causing reputational, privacy or other harm.

Appropriate labels may include:

• AI-Generated Illustration
• Conceptual Image — Not an Actual Photograph
• Synthetic Visual Representation
• AI-Assisted Artwork

Disclosure should appear in a location reasonably connected to the image.

A caption should not falsely identify a synthetic scene as an authentic record of an event.

11. Altered Photographs and Composite Images

Images may undergo routine technical adjustments, such as reasonable cropping, exposure correction and resizing.

However, alterations that materially change the factual significance of a photograph require additional scrutiny.

Examples include:

• Removing an individual from a factual news image.
• Adding objects that were not present.
• Changing the setting of an actual event.
• Combining unrelated photographs into a misleading composite.
• Altering a person’s appearance in a way that affects the meaning of reporting.
• Removing identifying or contextual elements that are significant to the story.

Material modifications should not be concealed when they could mislead readers.

A manipulated image should not be presented as an unaltered original.

12. Deepfakes and Synthetic Impersonation

Deepfake technologies may create realistic representations of people appearing to say or do things they never said or did.

VNN must not knowingly present synthetic impersonations as authentic evidence.

Particular caution is required for synthetic material involving:

• Political figures.
• Public officials.
• Criminal allegations.
• Public emergencies.
• Elections.
• Financial announcements.
• Healthcare information.
• Intimate or sexualised imagery.
• People experiencing vulnerability.

Where synthetic material is itself the subject of legitimate reporting, the article should explain its nature and avoid unnecessarily amplifying deception or harm.

13. Synthetic Audio and Voice Cloning

AI-generated voice content must not be represented as genuine speech by a real person unless the audio accurately reproduces authorised authentic content and its nature is not misleading.

Any synthetic narration or voice recreation should be assessed for:

• Consent and permissions.
• Authenticity.
• Potential impersonation.
• Risk of misleading attribution.
• Copyright and related rights.
• Required disclosures.

Readers or listeners should not be led to believe that a real interview occurred when the recording was artificially generated.

14. AI Disclosure Requirements

VNN supports proportionate disclosure of AI involvement.

Not every minor use of spelling correction or formatting assistance requires an article-level disclosure.

However, disclosure should be provided where AI involvement is material to a reasonable reader’s understanding of the content’s origin, authenticity or reliability.

Circumstances That May Require Disclosure

• Substantial AI-generated editorial text.
• Synthetic images depicting real-world scenarios.
• AI-generated or materially modified video.
• Synthetic voice recordings.
• Automated explanatory content presented to readers.
• Materially AI-transformed source material.
• AI-generated reconstructions of actual events.

Example Disclosure — Editorial Assistance

“Artificial intelligence tools assisted with preparation of this article. The published material was reviewed and fact-checked by VNN editorial personnel.”

This wording must be used only where the stated human review and fact-checking actually occurred.

Example Disclosure — Synthetic Image

“The accompanying image is AI-generated and is provided for illustrative purposes. It does not depict an authentic photograph of the reported event.”

Example Disclosure — Automated Summary

“This summary was generated using automated technology. Material facts should be assessed against the attributed source reporting.”

Disclosures should be accurate, understandable and proportionate.

15. Privacy and Confidential Information

AI systems may process information through external service providers.

Editorial personnel must not assume that information entered into an AI tool remains confidential or is excluded from further processing.

Before submitting confidential material, VNN should consider:

1. The sensitivity of the information.
2. Whether personal information is involved.
3. Whether the tool retains submitted content.
4. Whether information may be used for model training.
5. Whether data is transferred internationally.
6. Whether access controls are appropriate.
7. Applicable contractual and legal obligations.
8. Whether anonymisation or another safer approach is possible.

Confidential source identities, unpublished investigations and sensitive personal records require heightened protection.

16. Copyright and Intellectual Property

AI use does not remove copyright responsibilities.

Editorial personnel should ensure that third-party source material, photographs, footage and other protected works are handled lawfully.

AI-generated content should not be assumed to be free from intellectual property risks.

Where commercially licensed AI tools are used, relevant contractual conditions should be reviewed.

VNN must not intentionally use AI to reproduce substantial protected content without a lawful basis.

Original creators and reporting organisations should receive appropriate attribution.

17. AI-Assisted Research and Academic Material

AI can assist with locating research questions and organising source material.

However, research findings should be checked against the original study or another reliable publication.

Particular care is required with:

• Medical research.
• Clinical recommendations.
• Scientific statistics.
• Legal decisions.
• Economic forecasts.
• Election information.
• Technical safety claims.

AI-generated summaries may omit important methodological limitations.

A source should not be cited unless it actually exists and supports the statement being made.

18. Automated Content Aggregation

VNN may use automated systems to identify or organise externally published news.

Aggregation systems must be configured and supervised with appropriate consideration for:

• Accurate source identification.
• Copyright and licensing.
• Duplicate reporting.
• Factual reliability.
• Outdated content.
• Incorrect publication dates.
• Misleading article summaries.
• Commercial and sponsored content.
• Correction updates from original publishers.

Automated aggregation must not misrepresent third-party journalism as original VNN reporting.

19. Automated Publication and Editorial Controls

Where VNN adopts automated publication workflows, the organisation should establish controls appropriate to the editorial risk.

High-impact material should not bypass necessary verification merely because the publishing system is technically capable of immediate release.

Appropriate controls may include editorial approval, trusted-source restrictions, publication thresholds, monitoring, audit records and emergency withdrawal procedures.

These controls must reflect actual system capabilities.

VNN should not claim that every automated article has been individually checked by a journalist unless that is true.

20. AI in Opinion and Commentary

AI may assist with structuring or editing opinion material.

However, a real individual must not be falsely identified as holding opinions generated without their knowledge or authorisation.

Guest contributors should remain accountable for factual assertions and the views published under their names.

Substantive AI authorship or transformation should be disclosed when necessary to avoid misleading readers.

21. AI-Related Errors and Corrections

If AI-assisted material contains a significant error, VNN should apply its ordinary corrections procedures.

Where necessary, the review should consider whether the AI system produced additional related inaccuracies.

Appropriate responses may include:

• Correcting factual errors.
• Adding missing source attribution.
• Updating an AI disclosure.
• Removing misleading synthetic imagery.
• Retracting seriously defective content.
• Reviewing the relevant automated workflow.

An AI system’s involvement does not reduce VNN’s editorial responsibility.

22. Bias, Fairness and Representation

AI systems may reproduce biases present in their training data or input material.

Editorial personnel should assess whether AI-assisted content creates inaccurate, discriminatory or misleading representations.

Particular care is required when reporting on vulnerable populations, contested social issues and historically marginalised communities.

AI-generated conclusions should not be accepted merely because they are expressed in apparently neutral language.

23. Oversight and Responsibility

The publisher or designated editorial lead is responsible for approving material AI-related editorial practices.

Contributors should understand the limitations of the tools they use.

Where a material AI-related failure is identified, the relevant workflow should be reviewed.

New technologies should be assessed before being used for high-risk editorial functions.

24. Reader Questions About AI

Readers may request clarification regarding material AI involvement in VNN content.

Editorial Contact

Where appropriate, VNN should explain the nature of the technology’s contribution without disclosing confidential information, security-sensitive details or protected proprietary material.

A request for information does not create an automatic entitlement to internal prompts, confidential source records or third-party software documentation.

25. Policy Review

This policy should be reviewed at least annually and following material changes to AI technology, applicable law or VNN’s publishing systems.

26. Our Final Commitment

VNN embraces innovation that improves journalism while recognising that technology can amplify errors as easily as it can improve efficiency.

Our responsibility is to ensure that efficiency never becomes an excuse for fabrication, misleading presentation or inadequate verification.

Artificial intelligence can assist us in telling a story. It must never be allowed to invent the truth of that story.