How Bay City News Foundation builds trust through transparent AI experimentation

Published on May 18, 2026May 18, 2026 by Lynn Walsh | Reading Time: 6 minutes

When Bay City News Foundation began experimenting with AI, they knew transparency would be critical to maintaining audience trust. The newsroom spent months exploring ways to disclose its use of AI during a Trusting News cohort and has decided to take a bold approach by inviting readers into their experiments, explaining what worked, what didn’t and what it all means for the future of journalism.

Through a mix of news stories, explainer articles, podcasts, experimental tools and open calls for feedback, Bay City News created a living archive of their AI journey. On their “Experiments with AI” page, users can find:

• A public AI use policy outlining ethical guidelines
• Deep dives into their AI-driven experiments, including:
• Producing a podcast from written reporting using AI-generated scripts and audio
• Automating election results reporting, complete with a playbook for other newsrooms
• Experimenting with AI-generated cartoons
• A section featuring news stories about AI
• A survey requesting feedback to include the audience directly in the process

“We want readers to understand more about the power and pitfalls of AI when used to find information,” Bay City News staff said. “Part of it has to do with media and AI literacy; part of it has to do with greater transparency about how we are using AI.”

In an open letter to readers, the newsroom explained AI is already embedded in tools we use every day — from map apps to predictive text.

“Many of these tools are so naturalized into our daily routines that it is easy to forget that they use AI,” Bay City News staff said. “Instead, when we hear the term ‘Artificial Intelligence’ our minds tend to leap to the latest and least known examples of this technology. Gaining a better understanding of the full spectrum and potential of AI applications will help demystify this technology, and hopefully give us a better grasp on how to use it and what to avoid.”

As they began experimenting with AI, they decided against keeping AI decisions behind closed doors. Instead, Bay City News created a “sandbox” where reporters could test new tools and openly share their experiences. By being transparent about their experience and sharing it alongside the finished work, the newsroom gave readers a clear picture of what was produced by humans, where AI assisted and why those choices were made.

Impact and effectiveness

According to the newsroom, its strategy focused on transparency has earned positive feedback from users. While some expressed concerns about losing the “human touch” in certain workflows (for example, an AI-voice trained on a human editor who reads the news), many appreciated the newsroom’s decision to focus human effort where it matters most: more in-depth reporting and connecting with communities.

“The biggest success of our approach, we think, is providing narratives alongside each experiment we do,” Bay City News staff said. “It is the full disclosure about what tools worked and which applications failed that shows we are not afraid to experiment — but also that we know that sometimes experiments never become products ready for publication.”

The newsroom also received positive feedback and an award from the journalism industry for its AI Elections Project, which automated significant portions of their 2024 election results. They also created a playbook for other newsrooms looking to replicate the model.

Producing the content

The team describes the rapid evolution of AI as a “sink or swim” moment for journalism and said while adopting AI has been time-consuming and, at times, overwhelming, it has also opened the door to new storytelling formats, innovative products and more efficient workflows.

“AI has definitely made us think differently about our interactions with it and our strategy for using it,” Bay City News staff said. “One of our top takeaways is that AI is most effective when we understand the limitations and capabilities of the various tools on the market and know how to piece them together to get the product we want.”

Newsroom leaders said staff reactions to using AI are mixed. Some staff members are enthusiastic about experimenting with new tools and exploring innovative ways to produce and share information. Others are more cautious, raising concerns about potential mistakes and hallucinations.

To address these concerns, the newsroom has spent significant time discussing how to use AI responsibly and ethically. A key part of this process has been transparency. One important step, the staff said, was being clear about what they do and don’t allow.

“Creativity comes easily to us, and AI helps us operate efficiently enough to fulfill those ideas,” Bay City News staff said. “As with anything new, there’s a learning curve, and we have learned to pivot and try again. Although the projects we use AI for are all different, there are lessons from working with AI that we can apply across different initiatives.”

A Q&A with Aly Brown, Chloe Lee Rowlands, Katherine Rowlands and Ciara Zavala, the journalists behind the work

Read the Q&A with Aly Brown, Chloe Lee Rowlands, Katherine Rowlands and Ciara Zavala below to learn more about the process the team used to create this content and what the response from the community and newsroom has been. The Q&A has been lightly edited for punctuation and grammar.

Trusting News would like to thank Aly Brown, Chloe Lee Rowlands, Katherine Rowlands and Ciara Zavala and the entire Bay City News Foundation team for their time, effort and dedication to building trust with their community through transparency and engagement strategies like these. Through this work, which can be found at LocalNewsMatters.org, we are able to learn more about what works best to build trust with the public and then share it with the journalism community. If you are experimenting with building trust, let us know here.

Can you describe how long producing this content took?

Here’s a graphic we created for our RJI playbook where we wrote about our entire AI semi-automated election process. Although, the graphic shows that we spent more time this year with the use of AI versus last year, we set the foundation and heavy-lifting structure that we don’t need to recreate in the next election.

Would you use this trust-building strategy or content again? If so, would you do anything differently?

Yes, we will continue to use a trust-building strategy because, as media consumers ourselves, we understand the distrust surrounding AI use. A simple label like, “AI was used in the creation of this content,” does little to assuage a person’s concerns over how exactly AI has been used. So, we’ve done the complete opposite and brought the reader into an AI experiment that other media companies might treat as an internal discussion.

I think we can work on our disclosure language and be even more transparent. Although we believe that people need to be aware of our AI use, after being part of the Trusting News cohort and reviewing the survey results, we were most concerned with people believing that AI use meant a newsroom is lazy. In reality, we’re utilizing AI so that we can survive as a small newsroom with limited resources that can’t fundraise or push out content at the same speed as a bigger newsroom.

How would you rate the difficulty of producing this content?

I would rate the difficulty of producing AI content as a 6 to 8 out of 10, depending on the type of content. If a project is code-based and has a troubleshooting component, then it becomes more difficult. If it’s a simple back-and-forth conversation with an AI bot to gather research or get steps for building something out, then it’s a bit easier.

How would you describe the user response to this content?

While we don’t utilize comments on our site, and therefore have not received user feedback that way, [but] we have spoken to a number of individuals in our community about this project — both the content and the use of AI.

Over the course of our experimentation with AI, including the projects that we did prior to this experiment, we have received overwhelmingly positive feedback about how our approach to AI prioritizes transparency, openness and building trust with our audience.

The feedback we received about this project indicated similar appreciation for those aspects of our approach — readers and peers in the industry value that we share not only the in-depth details of what AI tools we use, how we use them and what that process looks like, but also that we openly point out the roadblocks and internal conflict we face during these experiments. Specifically, publishing our road map for this project and sharing openly about the ways that our early efforts for this project “failed” (aka our first attempt produced something boring and not worth the effort, and we told our audience that outright) and how we regrouped to change our strategy accordingly in order to iterate towards a more successful product.

However, while our transparency around this project received nothing but positive reactions, the specific use of AI was more complicated for people. This project pushed the limits on what we have previously done with AI.

Previously, our uses of AI have primarily been as a tool to support our journalistic efforts and we have strayed away from using AI to generate content. This experiment walked that line — while all the reporting was done by our human reporters and edited by our human editors, we did use AI tools to “translate” that written copy into a podcast script and then used AI tools to turn that script into audio content.

For some individuals both inside and outside of our newsroom, this use verged into the territory of using generative AI in the place of humans and left them feeling conflicted. The beauty of our approach is that we share those same conflicts and ask ourselves those same questions when it comes to the use of AI. The truth of the matter is that many out there are using AI in these exact ways to produce content like this, but without the transparency and reflection that we seek to prioritize and share.

We want to bring the reality of these tools into the public and offer information and critical reflection on them. By experimenting with AI and sharing those experiments transparently, we seek to tackle those questions and conflicts head-on and to share that with our audience so that it becomes an educational opportunity to understand AI on a deeper level and think about how we can approach it ethically.

---

At Trusting News, we learn how people decide what news to trust and turn that knowledge into actionable strategies for journalists. We train and empower journalists to take responsibility for demonstrating credibility and actively earning trust through transparency and engagement. Learn more about our work, vision and team. Subscribe to our Trust Tips newsletter. Follow us on Twitter, BlueSky and LinkedIn.