Leading with transparency and accountability when using AI

By the CMA Media Ethics Committee

The CMA Fair Publishing Practices Code, now including a section on artificial intelligence and ethics, urges CMA members to “strive to be transparent and accountable,” and to “clearly disclose when AI is used in generating editorial or creative content.”

Members of the Catholic Media Association believe communication is not merely the transmission of data, but an avenue for encounter that promotes the common good.

Generative artificial intelligence is reshaping how we communicate. Through these changes, we remain committed to protecting the dignity of the human person.

In Magnifica Humanitas, Pope Leo XIV calls us to a rigorous commitment to “transparency.”

Transparency helps ensure that technology serves humanity while we safeguard the human vocation.

The following practical guidelines for disclosure are not meant as a rigid checklist or a mere mechanism for institutional protection, but as a means to serve the truth and maintain the sacred trust we have built over the years with our audiences.

They are the fruit of discussions among those on the Media Ethics Committee, who have heard CMA members voice concerns on issues related to transparency and accountability, such as:

• When is disclosure required?
• What ethical boundaries should not be crossed?
• What should you disclose?

A 2025 article from the American Journalism Project highlights how major news organizations have adopted different approaches to AI.

Here are two examples showing a clear divergence in outlook to and strategy for AI in a newsroom:

• Associated Press has adopted very strict standards against AI-generated content, and why.
• Reuters has a more permissive policy.

Adopting guidelines that clarify when AI tools have been used in content creation or for significant editorial assistance is part of building transparency and public trust.

But what might that disclosure look like? To illustrate possible approaches, the Media Ethics Committee generated sample disclosure language using ChatGPT and then substantially revised it.

Here are a few suggestions:

1. Use simple disclosure statements that note when AI plays a significant role in content creation.

   These statements can be placed at the end of articles, in video descriptions or in the credits, such as:

   • “This article was developed with the assistance of AI tools and reviewed, edited and approved by our editorial team.”
   • “This image was generated using artificial intelligence.”
   • “Illustration generated by AI with direction from” (insert your news organization).
   • "AI-assisted editing tools were used in the production of this video."

   Note: Disclosure does not eliminate the need for editorial judgment, factchecking and human accountability.

2. Distinguish in what stage of production AI assistance/generation was used: in the editing phase, grammar, transcribing or translation… with human oversight; in the drafting phase, structuring or condensing content; in the research phase, brainstorming. For example:
   “AI-assisted editing” or “AI-generated content.”

   Mark Greenblatt, professor at Arizona State University and workshop presenter on AI and investigative reporting at the 2025 Catholic Media Conference, offered examples of “transparency language” that has the added benefit of indicating to our audiences that human dignity remains the center and animating force of our work:

   • “This graphic was generated using an AI tool and reviewed by our visuals editor.”
   • “This visual was AI-generated based on our reporting; it does not depict a real person or event.”
   • “Portions of this article were drafted using an AI language model and subsequently edited and fact-checked by our staff.”
   • “AI tools were used to assist in writing this summary, with final content reviewed by a human editor.”
3. Be especially transparent with visual media since AI-generated images and videos can be mistaken for real photographs or real people, places and events.
   For example: “This image is an artistic rendering created with AI and does not depict a real photograph.”
4. Create and publish a short and accessible AI Usage Policy on your website explaining: what kinds of AI tools are used; in what contexts (writing, editing, illustrations, etc.) were AI tools used; and the organization's commitment to human oversight and ethical standards.

As creating a media organization’s own accurate disclosure statement may seem challenging, designers, researchers and engineers at IBM Research developed the AI Attribution Toolkit.

The toolkit guides the user through four multiple-choice questions followed by one last fill-in-the-blank query to create an attribution statement specific to the media organization:

1. How much did AI contribute?
2. What did AI contribute?
3. Who initiated the work?
4. Was AI content reviewed by a human?
5. What AI “model or application” was used?

“An attribution statement identifies not only the presence of AI involvement, but also how AI was used. This approach makes important distinctions between different types and amounts of AI contributions, allowing creators to maintain ownership over co-created work and consumers to calibrate their trust,” according to the toolkit.

These practical measures can amplify some guiding principles as our members navigate the complex and rapidly changing digital terrain of an AI world while prioritizing the common good and the human storyteller.