How We Use AI in Our Reporting

Artificial intelligence tools are changing how newsrooms work, allowing us to be more efficient and opening up new doors to larger and more complex stories and investigations.

We use AI tools to assist with some parts of our jobs, but only our journalists do the reporting. A human writes every story we publish, a human checks every fact and a human makes every editorial decision before anything is published.

We will always transparently disclose how AI is used in our reporting process and how that work has been reviewed for accuracy and fairness by a human.

Our AI use policy is governed by our internal AI Working Group, composed of volunteers from across our organization representing our technology, editorial and business operations teams.

What We Use AI For

We use AI tools as assistants for certain tasks in our newsroom, in much the same way we use spreadsheets, databases and other software. Here are some examples:

1. Accessibility. We use AI tools to increase access to our reporting. This includes AI-generated audio articles, and may include other accessibility features such as AI-generated video captions and translations of our articles into other languages.
2. Research and data analysis. When we obtain large volumes of public records, AI can help us sort and process data faster than we could by hand. Our reporters always verify what the AI produces by reviewing it against the original records.
3. Transcription. AI tools can transcribe recorded interviews and public meetings, saving hours of manual work. Our reporters always check the transcript against the original audio before quoting anyone.

What We Do Not Use AI For

We have firm limits on how AI is used in our newsroom:

1. AI does not write our stories. Any article, newsletter, or other published text that carries a byline is written by our journalists and edited by a human editor.
2. AI does not make editorial decisions. What we cover, how we cover it and what we publish are decisions made by our journalists and editors.
3. AI does not create our photographs or videos. We do not use AI to generate, alter or enhance news photographs or video.
4. AI is not a source. We do not treat information generated by AI as factual. Everything AI produces is treated as unverified material that must be checked against real sources.

How We Check AI’s Work

We follow a simple principle called “human in the loop.” This means a human initiates the work, AI assists in the middle and a human reviews and finalizes the work before it reaches you. AI output goes through the same editorial process as any other material in our newsroom. We fact-check it, we question it and we hold it to the same standards we hold ourselves.

How We Protect Your Privacy and Our Sources

We do not enter confidential source information, unpublished reporting or private reader data into third-party AI tools unless the tools have been vetted and found to adhere to strict data protection policies. When we work with sensitive material, we use tools and methods that keep that information within our control.

How We Tell You About It

When AI plays a role in our reporting, we include a disclosure in the story explaining what was done, why it was done and how it was verified. You may have arrived at this page from one of those notes.

We do not disclose the use of routine AI-powered features built into common software such as spell-check, auto-complete or some search tools. We focus our disclosures on uses that materially contribute to the reporting.

This Policy Will Evolve

AI technology is changing rapidly and so is our understanding of how to use it well. We review our AI policy regularly and will update this page as our practices develop.

Questions or Feedback

We want to hear from you. If you have questions about how we use AI, concerns about something you’ve seen in our reporting, or thoughts on how we should approach this technology, please reach out to our AI Working Group at ai@deepsouthtoday.org.