The headline image was generated by artificial intelligence.

The Community Media Network, which operates Radio Al-Balad and AmmanNet, has added a set of new provisions to its publicly stated editorial policies governing the use of artificial intelligence tools in journalistic work. The move is aimed at making use of emerging technologies in newsrooms while preserving human accountability, accuracy, editorial independence, source protection, and privacy.

The new policy regulates the use of artificial intelligence at various stages of journalistic work, from preliminary research and interview transcription to translation, editing, headline suggestions, and data analysis. It also establishes clear restrictions on the use of AI in producing news reports, images, videos, and audio recordings, as well as in dealing with misleading content and deepfakes.

The policy states that artificial intelligence can serve as an assistive tool for journalists, but it is not considered a journalist or a source of information and cannot replace human editorial judgment. Full responsibility for any material published or broadcast in the name of Radio Al-Balad or AmmanNet remains with the journalist, editor, and editor-in-chief.

Mohammad Al-Ersan, Editor-in-Chief of Radio Al-Balad, said artificial intelligence has become part of the transformation taking place in journalism, noting that it is now being used for a range of tasks that previously required considerable time and effort in newsrooms.

Al-Ersan said some of the most practical uses of these tools include transcribing interviews and audio recordings, assisting with editing, suggesting headlines and editorial angles, producing short-form video clips, and supporting certain design and promotional tasks.

He said these tools have significantly reduced the time required for some tasks, but practical experience has also revealed errors in transcription and information processing. This has made human review an essential requirement, rather than an optional step, before any AI-generated output is used for journalistic purposes.

Al-Ersan stressed that the rapid development of artificial intelligence technologies requires media organizations to establish clear rules defining how these tools may be used, what uses should be prohibited, and who bears responsibility for published material.

He added that any policy governing artificial intelligence should be rooted in fundamental journalistic principles, particularly accuracy, credibility, and the prevention of misinformation, while keeping pace with technological developments without allowing technology to take precedence over professional responsibility.

What Are the Main New Provisions?

The policy introduced by Radio Al-Balad and AmmanNet includes a broad set of safeguards. Among the most important is the principle that the use of artificial intelligence must serve the organizations’ core mission, namely providing accurate, independent, and reliable information in the public interest. The policy also makes clear that human beings must retain final decision-making authority at every editorial stage where AI technologies are used.

The policy permits the use of artificial intelligence tools for preliminary research, summarizing documents, transcribing recordings and interviews, assisting with translation, proofreading, suggesting headlines, introductions and summaries, organizing information, analyzing large datasets, and identifying patterns or inconsistencies that may warrant further journalistic investigation. However, the use of these tools must never eliminate or undermine the processes of information gathering and verification.

The policy emphasizes that AI-generated outputs must be treated as preliminary material requiring verification, rather than as established facts ready for publication. Every piece of information, figure, name, statement, or event provided by an AI system must therefore be verified against original, credible, and independent sources.

Prohibiting Fabricated Sources and Statements

One of the policy’s key provisions prohibits the use of artificial intelligence to generate sources, statements, testimonies, or data that do not exist in reality.

The policy also prohibits publishing text generated entirely by AI as original journalistic work without human review, editing, and verification. Journalists must be able to explain the essential information contained in published material, identify its sources, and justify its inclusion.

Particular attention is required when reporting allegations, sensitive cases, or information that could affect an individual’s reputation or safety. In such cases, the policy requires a higher level of human review and verification.

Images, Video, and Deepfakes

The policy imposes strict restrictions on the handling of images, audio recordings, and video footage, recognizing their role as forms of journalistic evidence.

It prohibits adding or removing substantive elements from a news image, audio recording, or video in a way that alters the reality represented by the material. It also prohibits creating synthetic content that could lead audiences to believe they are viewing authentic documentation of an event that never occurred.

The policy likewise prohibits the use of voice or facial cloning technologies to attribute statements to a person who did not make them or to depict them as appearing in material in which they did not actually appear. At the same time, AI-generated material may be used as part of a journalistic story about artificial intelligence or for explanatory or artistic purposes, provided that its synthetic nature is clear to the audience and it is not presented as authentic documentation.

Material suspected of having been generated or altered using artificial intelligence must be treated as content requiring verification. Journalists should seek to identify its original source, verify its date and context, and compare it with independent sources whenever possible. If its authenticity cannot be established, it must not be presented to the public as genuine.

Protecting Sources and Sensitive Data

The policy prohibits entering confidential or sensitive information into public or institutionally unapproved artificial intelligence tools. This includes the names of confidential sources, telephone numbers, personal data, unpublished documents, sensitive medical, financial, and legal information, passwords, and any material that could reveal the identity of a confidential source.

When using an external tool to process journalistic material, journalists are required, to the extent possible, to review the tool’s data-handling and privacy terms before uploading any material.

Transparency with Audiences

The policy establishes transparency as a key principle in dealing with audiences. Radio Al-Balad and AmmanNet are committed to disclosing the substantive use of artificial intelligence in the production of journalistic content.

The use of AI for routine technical tasks, such as spelling and grammar checks or organizing notes, does not in itself require disclosure unless it has a substantive impact on the published content. When artificial intelligence plays a significant role in producing or modifying published material, however, the audience must be informed in a clear and understandable manner.

Audiences must also be made aware when they are viewing or hearing synthetic material that indirectly represents a real person or event.

Editorial Independence and Algorithmic Bias

The policy prohibits the use of artificial intelligence in ways that would allow advertisers, funders, political actors, government bodies, or commercial entities to influence the selection, ranking, framing, or wording of news stories, or determine which individuals are interviewed.

It also makes clear that decisions concerning news priorities remain subject to human editorial judgment. Such decisions cannot be delegated to an algorithm that independently determines whether a particular issue deserves coverage.

The policy also addresses the risk of algorithmic bias, requiring journalists to scrutinize AI-generated outputs, particularly when dealing with women, children, minorities, marginalized groups, geographic communities, and people who may be vulnerable to discrimination. It also prohibits relying on classifications or descriptions generated by AI about individuals or groups without verification.

Artificial Intelligence in Investigative Journalism

The policy extends beyond daily news reporting to investigative journalism, allowing AI tools to be used to analyze large volumes of documents and data in order to identify patterns, inconsistencies, or information that may warrant investigation.

At the same time, the policy draws a clear line around such use. The identification of a pattern or relationship in data by an AI system does not, by itself, constitute journalistic evidence. Journalists must return to the original documents and sources and verify the findings before using them in an investigation or report.

Corrections and Training

If the use of artificial intelligence results in the publication of incorrect information, responsibility for correcting the error remains with the organization, the journalist, and the responsible editor. AI-related errors are subject to the same correction rules established under Radio Al-Balad and AmmanNet’s existing editorial policy.

The policy also calls for the organization to identify the tools and platforms approved for journalistic work and to review the approved list regularly. It further requires journalists, editors, and production staff to receive training on the risks associated with artificial intelligence, including hallucinations, bias, deepfakes, data privacy, intellectual property rights, and methods for verifying synthetic content.

The move comes as international news organizations are developing increasingly detailed rules governing the use of artificial intelligence in newsrooms. On July 23, 2026, the Associated Press updated its newsroom standards on artificial intelligence, emphasizing that the technology can assist journalists with specific tasks, but that editorial judgment, verification, and accountability remain the responsibility of journalists. Permitted uses include preliminary research, document summarization, transcription, translation, and headline suggestions, with AI outputs subject to review and editing before publication.

The Associated Press also treats AI-generated outputs as unverified material requiring editorial assessment, warns against entering confidential or sensitive information into AI tools, and places restrictions on the use of artificial intelligence in news photography and visual content.

Reuters, meanwhile, has said that artificial intelligence is being used in various aspects of the news production process, while stressing the importance of disclosure when content has been substantially or entirely generated by AI and the need to maintain editorial safeguards.

The new policy adopted by Radio Al-Balad and AmmanNet therefore does not treat artificial intelligence as either a prohibited technology or a substitute for journalists. Instead, it recognizes AI as a tool that can be used within clearly defined boundaries. The central principle behind the new provisions is that greater speed or increased output must never come at the expense of verification, accuracy, source protection, or editorial independence.

The policy sums up its approach in a clear rule: when a journalist cannot verify content produced by artificial intelligence, that content must not be published. If the use of the technology could cause irreparable harm or mislead the public about a fundamental fact, it must not be used.

Policy on the Use of Artificial Intelligence at Radio Al-Balad and AmmanNet

I. General Principles

The Community Media Network recognizes that artificial intelligence technologies have become part of the modern media environment and may be used to enhance journalistic work, improve access to information, and increase production efficiency, provided that their use does not compromise the professional and ethical principles set out in this policy.

The use of artificial intelligence must serve the core mission of Radio Al-Balad and AmmanNet, which is to provide accurate, independent, and reliable information that serves the public and the public interest.

Artificial intelligence shall not be treated as a journalist, a source of information, or a substitute for human editorial judgment.

Full responsibility for any material published or broadcast in the name of Radio Al-Balad or AmmanNet rests with the journalist, editor, and editor-in-chief. Such responsibility cannot be transferred to the tool or the company that developed the artificial intelligence system.

II. Human Responsibility

Human beings must retain final decision-making authority at every editorial stage in which artificial intelligence is used.

No material prepared or assisted by an artificial intelligence system may be published or broadcast without appropriate human review based on the nature of the material.

Journalists must not regard an AI system’s response as a reliable source in its own right, even if the response appears accurate or contains source names, links, or figures.

AI-generated outputs must be treated as preliminary material requiring verification and assessment, rather than as facts ready for publication.

III. Verification of Information

Every piece of information, figure, name, statement, or event provided by an artificial intelligence system must be verified against original or credible and independent sources before publication.

The policy prohibits reliance on what is commonly known as AI "hallucinations", including fabricated information, nonexistent sources, and quotations that were never made by the individuals to whom they are attributed.

Artificial intelligence must not be used to generate sources, statements, testimonies, or data that do not exist in reality.

If information suggested by an AI system cannot be verified, it must not be published.

IV. Use of Artificial Intelligence in News Production

Artificial intelligence tools approved by the organization may be used for:

• Preliminary research into potential topics or angles for coverage.
• Summarizing lengthy documents, provided that the original document is consulted for verification.
• Transcribing audio recordings and interviews.
• Assisting with translation, provided that the translation undergoes human review.
• Proofreading and preliminary editing.
• Suggesting headlines, introductions, or summaries.
• Organizing information and data.
• Assisting in the analysis of large datasets.
• Identifying patterns or inconsistencies that may warrant journalistic investigation.
• Assisting in preparing questions or follow-up angles.

The use of these tools must not eliminate the process of gathering and verifying information, nor reduce it to a level that compromises journalistic quality.

V. News Writing and Published Content

Text generated entirely by artificial intelligence must not be published as original journalistic work without review, editing, and verification by a journalist or editor within the organization.

Journalists must be able to explain every essential piece of information contained in published material, identify its source, and justify its inclusion.

Artificial intelligence must not be used to produce fabricated news, fill gaps in unavailable information, or reconstruct events that have not been verified.

Artificial intelligence must not independently determine whether a person or organization is guilty of, or responsible for, an offense or crime.

For reports involving allegations, sensitive cases, or information that could affect individuals’ reputations or safety, human review and verification must be subject to a higher level of scrutiny.

VI. Images, Video, and Audio

Images, audio recordings, and video footage are considered forms of journalistic evidence and are therefore subject to the highest standards of protection against manipulation.

Artificial intelligence must not be used to add or remove substantive elements from a news image, audio recording, or video in a way that changes the reality represented by the content.

No synthetic image, video, or audio recording may be created in a way that leads audiences to believe it is authentic documentation of an event that never occurred.

Voice or facial cloning technologies must not be used to attribute statements to a person who did not make them or to depict that person as appearing in material in which they did not actually appear.

Synthetic or AI-processed material may be used when it forms part of a journalistic story about artificial intelligence or serves an explanatory or artistic purpose, provided that its nature is clear to the audience and it is not presented as authentic documentation.

Any AI-generated image, video, or audio material whose editorial use is justified must carry a clear indication that it was generated using artificial intelligence.

VII. Deepfakes and Manipulated Content

Images, videos, and audio recordings suspected of being generated or altered using artificial intelligence must be treated as material requiring verification, rather than as reliable evidence.

Before publishing any such material, journalists must attempt to identify its original source, verify its date and context, and compare it with independent sources whenever possible.

If the journalist cannot verify the authenticity of the material, it must not be presented to the public as genuine.

When a fabricated or AI-generated item is covered in a news report or feature, it must be made clear to the audience that the material is fake, generated, or manipulated. The report should also explain why the material is journalistically significant.

VIII. Privacy and Source Protection

Confidential or sensitive information must not be entered into public or institutionally unapproved artificial intelligence tools.

This includes the names of confidential sources, telephone numbers, personal data, unpublished documents, medical information, financial information, passwords, sensitive legal information, and material that could reveal the identity of a confidential source.

No artificial intelligence tool may be used in a manner that puts source confidentiality at risk.

When using an external tool to process journalistic material, journalists must, to the extent possible, verify the tool’s data-handling and privacy terms before uploading the material.

IX. Intellectual Property Rights

Artificial intelligence must not be used as a means of stealing or reproducing journalistic, literary, or visual material owned by others.

The rights of authors, publishers, photographers, musicians, and media organizations must be respected when using AI-generated material or artificial intelligence tools.

AI-generated text, images, or recordings must not be presented as original work produced by another person.

X. Transparency with Audiences

Radio Al-Balad and AmmanNet are committed to transparency regarding the substantive use of artificial intelligence in the production of journalistic material.

Disclosure is not required for every minor technical use of artificial intelligence, such as spelling correction or the organization of notes, unless such use has a substantive impact on the published content.

When artificial intelligence plays a substantive role in producing or modifying published material, the audience must be informed clearly and understandably.

Audiences must be informed when they are viewing or hearing synthetic material that indirectly represents a real person or event.

XI. Editorial Independence

Artificial intelligence tools must not be used in a manner that allows an advertiser, funder, political, governmental, or commercial entity to influence the selection, ranking, wording, or framing of news, or determine which individuals are interviewed.

The organization must not use artificial intelligence to make editorial decisions in place of editors, such as determining solely on the basis of an algorithm that a particular issue does not warrant coverage.

News coverage priorities remain subject to human editorial judgment and the editorial policies of Radio Al-Balad and AmmanNet.

XII. Non-Discrimination and Algorithmic Bias

Journalists recognize that artificial intelligence systems may reflect biases present in the data on which they were trained or in the systems’ design.

AI-generated outputs must therefore be scrutinized, particularly when dealing with women, children, minorities, marginalized groups, geographic communities, and individuals who may be vulnerable to discrimination.

No classification or characterization generated by artificial intelligence about an individual or group may be adopted without verification.

Artificial intelligence must not be used to produce content that reinforces stereotypes, racism, sectarianism, regional discrimination, or discrimination on any other basis.

XIII. Artificial Intelligence in Interviews and Audience Engagement

An artificial intelligence system must not be presented to the public as a journalist or a real person.

If the organization uses chatbots or automated tools to interact with audiences, users must be clearly informed that they are interacting with an automated system.

An automated system must not present itself as a human source or give the impression that its responses represent the views of a journalist or editor within the organization.

A clear mechanism must be provided for referring sensitive matters, complaints, or important news information to a human staff member.

XIV. Use of Artificial Intelligence in Investigative Journalism

Artificial intelligence tools may be used to analyze large volumes of documents or data in order to identify patterns, inconsistencies, or information that may warrant investigation.

However, the identification by an AI system of a pattern or relationship between datasets does not, in itself, constitute journalistic evidence.

Journalists must return to the original documents and sources and verify the findings before using them in any investigation or report.

XV. Errors and Corrections

If the use of artificial intelligence results in the publication of incorrect information, responsibility for correcting the error rests with the organization, the journalist, and the responsible editor. The AI system or its developer must not be blamed publicly for the error.

Errors resulting from the use of artificial intelligence are subject to the correction procedures established under the editorial policy of Radio Al-Balad and AmmanNet, with the same speed and clarity applied to any other journalistic error.

If artificial intelligence played a substantive role in causing the error, this may be disclosed to the public when such disclosure is necessary to explain how the error occurred.

XVI. Approved Tools and Training

Management, in coordination with the editorial team, shall determine which tools and platforms may be used in journalistic work and shall review the approved list periodically.

Journalists, editors, and production staff must receive training on the risks associated with artificial intelligence, including hallucinations, bias, deepfakes, data privacy, intellectual property rights, and the verification of synthetic content.

The use of artificial intelligence must not weaken essential journalistic skills, particularly information gathering, interviewing, source verification, critical thinking, and editing.

XVII. Core Ethical Rule

When in doubt, priority must be given to truth, the public interest, people’s safety, source protection, and editorial independence.

If a journalist cannot verify content produced by artificial intelligence, it must not be published.

If the use of artificial intelligence could cause irreparable harm or mislead the public about a fundamental fact, its use must be avoided.

Speed, cost reduction, or increased production must never be used as grounds for overriding any of the professional standards established in this policy.