The National Press Foundation’s Principles in Using Generative AI

Businesses, nonprofit organizations and newsrooms alike are using generative AI, and NPF is no exception. While these tools can improve efficiency, we must understand their limits. NPF’s mission, to “make good journalists better,” is achieved by improving the knowledge and skills of journalists so they may better serve the public. It is with this in mind that we adopt a simple principle when it comes to AI: All NPF content originates with human intellect and is completed with human oversight. NPF staff will be transparent about the use of any gen AI tools and is ultimately responsible for the accuracy of all we produce. 

Definition

Gen AI is a type of artificial intelligence that can create new text, images, videos, music or code based on inputs and prompts. A primary type is Large Language Models (LLMs), such as ChatGPT, Gemini and Claude.

Accuracy – All information generated by AI requires human verification. 

Transparency 

When AI tools are used in the process of publishing public-facing content, NPF will disclose the name or type of tool/app used and the purpose. This information will be shared in cutline/credit lines or in italics below the body of text.

When AI tools are used internally to complete tasks that typically require human judgment and creativity, staff must inform their manager.

Generative AI tools we may use:

  • ChatGPT, Gemini or Claude
  • Rev for audio/video transcription
  • Opus for video clipping
  • YESEO for brainstorming headlines, subject lines, etc.

Tools we use that include AI features:

  • Airtable
  • Canva
  • Zoom

Generative AI tools we do not use:

  • We do not use Gen AI to generate illustrations, photos, videos or music.

We will continue to evaluate tools on a case-by-case basis and will update this statement with any notable changes. 

Privacy 

NPF staff will not enter personal, sensitive or identifying information about any NPF affiliates (staff, speakers, fellows, etc.) into any generative AI tools.

No confidential information may be entered into a Gen AI tool/LLM.

Experimentation

As a journalism training organization, NPF must keep abreast of how Gen AI is being or could be used in newsrooms. In order to do this, NPF staff will explore tools even if they are not used in publication. 

AI companies

We do not have a formal relationship with and have not received any funding from LLM companies. This will be updated if that changes.

In creating these guidelines, NPF referenced those put forward by other media organizations, including Poynter, NPR, The Guardian, Axios, PBS and Wikimedia. 

These guidelines will be adhered to by the entire staff, including any part-time or contract staff, and were reviewed by the NPF AI Working Group members of the NPF Board of Directors.