Hi all. Hope everyone had a safe, relaxing holiday weekend!

In our latest Exploration we’ve got:

💻 A look back at the last 6 months of PMI webinars
🤞 The results of the latest YouGov trust in media survey
😎 Vibe coding for Journalists
🤖 Publishers publishing for bots
👻 Writing with AI
🥵 AI’s water problem
🦆🦆➡️ And, finally, DuckDuckGo gets the nod from Google

I will be at the Public Media Growth Conference in Chicago, later this week, and specifically part of the panel discussion AI Made Practical: Real Talk About What's Worked and What Hasn't, Friday at 1pCT. This one is hosted by friend-of-PMI, Erica Osher, VP of AI at NPR, and I will tell you, from the planning session, that Amber Samdahl from PBS Wisconsin, and Tarik Moody from Radio Milwaukee have some good demos to show. Reach out if you’ll be in Chicago!

Then, on July 16 at 1pET, we have our next webinar, How Public Media Companies are Building AI Policies. One of our quarterly collaborations with Current, this time we’re circling back around to the topic of AI policies. Presenters from public media companies (including your friendly neighborhood newsletter author) will share their experiences developing AI policies from the ground up. We'll cover what drove each station to act, how they structured their policies, and the lessons that came out of the process. Plus, we’re going to share our policies directly, so you’ll have some takeaways to get you start on drafting or updating your own policy. You can register now!

And, speaking of webinars….

Halfway through 2026, it feels like a good time for a look back over the 6+ hours of conversations and presentations on innovation that we’ve pulled together thus far this year.

Our June webinar was a reprise of one of the best presentations we saw at the PBS Annual Meeting. In this webinar we go inside PBS News’ Reddit strategy, from posting news clips in topic-specific subreddits to building genuine relationships with subreddit moderators. Dan Cooney, Vanessa Dennis, and Hannah Grabenstein share what they’ve learned about building authentic community on Reddit, what the numbers look like, and where local stations might start.

In May, we asked our friends at Connecticut Public to reprise their highly insightful session from the PBS Technology Summit in AI for Public Media: A Case Study with Connecticut Public. What you’ll see is an approach that merges leadership in journalism, tech and code, and I don’t mind telling you I am collegially envious of what they’ve put together. Our panelists included Susan Bell, Brendan Foley, Rob G. and Jim Haddadin. Plus, this one features Lauren Komrosky, who was recently named CEO of CT Public. Seems clear to me…appear in a PMI webinar, get promoted! 👇

If you missed our April webinar collaboration with Current, What’s Working with Public Media Newsletters we’ve got that for you as well. Email newsletters have become a powerful tool for building loyal audiences and creating a direct line to the communities we serve. Unlike social platforms, your newsletter list belongs to you — no algorithm, no platform risk, just a direct connection to the people who want to hear from you. It’s my opinion that there is no future for public media if we do not win the battle of the inbox. Chanelle Berlin Johnson, Madeleine King, Brianna Lee, and Jenny Peek give you a good snapshot of what it takes to win. gives you a good snapshot of what it takes to win. 👇

In March we brought you Myths & Truths of Working with Influencers, featuring Matthew Baltzell, Heather Reese, and Michael Shea. Working with creators and influencers is becoming an increasingly important part of how public media organizations reach new audiences and explore different approaches to digital storytelling. In this webinar, colleagues from across the system share real-world examples of working with external influencers, developing internal talent, and creating content designed specifically for social platforms, all while staying grounded in public media values. 👇

And for those interested in this topic you’ll also want to review (or revisit) the American Press Institute’s guide to influencer collaborations, published last fall.

Rolling the clock back to February, we explored How NPR Is Using AI. In this conversation, Erica Osher, Tony Cavin and Sharahn Thomas from NPR discuss how their teams are testing and using AI in newsroom workflows, how they evaluate tools before adoption, and how standards and cross-team collaboration shape their approach. 👇

And then, for January, we offered up the second edition of our Future of Public Media webinar (which, we hope, will be an annual event going forward). Click this one and join panelists Steve Bass, Nathalie Hill, Erik Langner, and Loira Limbal for an open and candid conversation about where public media finds itself at the start of a new year shaped by uncertainty and change. 👇

Okay, on to the links.

Trust in Media 2026: Which news sources Americans use and trust (Taylor Orth & Carl Bialik - YouGov) - Somehow, year after year, I hear we're the most trusted brand in media but never hear that The Weather Channel is actually more trusted. I'm sure there is carefully worded framing that ensures public media's assertion of primacy is literally correct. But in the interest of being honest amongst ourselves, we should tip a hat to The Weather Channel, and also to BBC (#4) and NPR (#9).

A Tough Day for NPR (Charlie Warzel - The Atlantic) - Speaking of trust. 🙄 On the bright side, this wasn’t an AI-generated crisis.

Public Media Is All the Rage (Asad Muhammad - Color Congress) - This is a good piece that calls into question what is the "public" in public media. The straightforward answer is "the people." That's the point I see this making. But I think "publics" are something more politically relevant than just “the people.” I might write about that sometime down the road, but for now this piece will get you thinking.

EVENT: Emerging Tech & AI: Tools, Ethics & Smarter Workflows (Public Media Journalists Association) - Last week we connected with new friends at the PMJA and were pleased to find there’s a certain affinity between our mission and theirs. (Turns out, there’s also a coincidental affinity between our branding color schemes, so don’t get us confused. 😉) Their Director of Member Learning & Community, Aubrey Nagle, has curated a very interesting agenda, so I wanted to highlight their event, September 9, 10a-5p, in NYC. Here’s their description: “AI isn’t coming. It’s already in your newsroom. The question is whether you’re using it well. Every public media journalist is navigating the same moment: tools that can genuinely help sitting next to real risks, unclear policies, and a lot of noise about what AI means for the industry. Near + Now New York City cuts through that noise. Journalists, producers, editors, news leaders, and every role in between spend the day getting practical: what responsible use actually looks like, which tools are worth your time, how to build the policies your newsroom needs, and what happens when things go wrong. Hosted at Google’s New York offices through the Google News Initiative, this event includes hands-on time with tools and real case studies from newsrooms that have been figuring this out in the open.”

Public Media Bridge Fund Announces $125,000 Investment in Newsletter Platform Quoted Network (Public Media Bridge Fund) - For those of you doing newsletters (and you should all be doing newsletters) I think this system is worth a look. It solves the problem of keeping your newsletters feeling robust (which is often what causes delays in me releasing this newsletter). My understanding is that they make money by selling public media appropriate sponsorship into an add block that comes with the content embed (which stations can participate in as well).

Vibe coding for journalists: Build interactive stories without writing a single line of code (Kris Krüg - The Media Copilot) - I liked this line: "The article format has served us well. It’s not dead, but it is not the only option." We are moving toward a future where journalists (and eventually agents serving publications) can spin up micro-programs, apps or other pieces of software for individual stories. We're not there yet (definitely pay heed to Krüg's "three high-level no-no's"), but if you've been curious what the fuss is about "vibe coding" this piece is a good primer. And if you AI policy doesn't address vibe coding, you should put some guardrails around that soon.

Unconventional (Gina Chua - (Re)Structured News) - This piece elevates some good points about the incentives humans have, as part of a journalistic enterprise, that AI doesn't have. Accountability is a product of norms, and we can't assume agents subscribe to those same norms (not yet at least). I don't know that anyone reading this newsletter is going to be advocating for an automated newsroom, but this piece can help you put a frame around those arguments against fully automated journalism.

We Asked the Future of Truth Author to Explain How He Used AI. It Didn’t Go Well (Kate Knibbs - Wired) - the headline of this piece hints at a "gotcha," but really, there isn't one. This piece is a class in how not to talk about AI use in creation, but it's also interesting to read the comments which mix a goodly amount of anti-AI vitriol with readers taking Wired to task (as a future-leaning, tech-oriented publication) for being so seemingly biased against AI use. What doesn't help any of this is the subject of the piece. He heads the Sustainable Media Center, whose mission I support and to whose Substack (largely penned by him) I recently unsubscribed because of its history of being holier-than-thou when it comes to tech issues. (I'm not saying he's wrong, but methinks he doth clutch his pearls too often.) So, if you have the larger context, reading this piece can be a bit of a tug-of-war between cringe and schadenfreude.

How Time and others are rebuilding parts of the web for AI agents (Sara Guaglione - Digiday) - We've been talking here about the need to "write" content for bots as much (or instead of) humans. This piece presents some very interesting tactics for consideration. It's worth discussing this with whomever manages your web experience.

The volunteer Wikipedia army protecting against AI slop (Ananya Bhattacharya - Rest of the World) - Here is basically my argument for why we should allow public media content to be ingested by AI models: '“Since Wikipedia is built on verified, reliable sources, it is actually a good thing that AI tools scrape our data rather than non-credible sources,” he said. “I want to ensure the ‘brain’ of the internet is trained on quality content.”'

Writing with AI (Jim VandeHei - Axios) - I found this piece quite a useful way to approach writing with AI, especially how he set up his tools with specific instructions. And I'm going to appropriate the concept of Lazy AI, which he categorically (and I think rightly) rejects at the start of this piece. I use AI for copy editing (typos, mainly, occasionally tightening up some language) in this newsletter, but I don't use it to write my columns. That's because, for me, writing is in a feedback loop with my thinking. And I figure, you've subscribed to this for my thinking. But as I've been working on an update to our AI policy, I've used Gemini and ChatGPT extensively (as I did in the first draft). In that case, the final output will be the product of my thinking as well as a host of others. There will be so many fingerprints on it, that all that matters is how effective it is, not who dropped in which pithy analogy or Oxford comma.

Your phone screen doesn’t have the same color range as the human eye – and AI widens the gap between digital images and the real thing (Douglas Goodwin - The Conversation) - I know that many of you (especially the video editors in the crowd) know that screens only show limited, 'good enough' color range. But I had never taken the next step by thinking about how the 'color bias' from the billions of images used to train AI would only be offering a narrow (color) vision of the world. Of course, that's also going to impact folks (like the author) with color blindness. Know your tools.

Understanding the Luddites in the age of AI (Brian Merchant - Blood in the Machine) - I’ve referenced Merchant's work before, both here and in live presentations, and his book on the Luddites (from which his Substack takes its name) would be a good beach read this summer for the historically, or technologically inclined. This post will give you a good taste of Merchant's style, thesis, and subject matter.

AI has a water problem — Google thinks it has a fix (Lauren Feiner - The Verge) - I know that the environmental concerns around AI are top of mind for you, members of your teams, or your colleagues, so I'm going to try and include more about this in the future (both stories that are critical and stories that set the record straight on some misunderstandings).

The Rise of the AI Censors (Hunter Walker - Talking Points Memo) - While I wouldn't take this report as the definitive word on the subject, for those public and non-profit media organizations fighting to establish a presence on TikTok this could be good intel. '[Ryan] Broderick [who runs the Garbage Day newsletter] said that, based on his conversations, he believes accounts with six-figure followings have largely been able to dodge the aggressive AI censorship. “My read on this is, like, if you can fight through the kind of brutal establishment of your account and you can get a big audience, TikTok will just sort of give you carte blanche,” Broderick explained.'

Meta's new app 'Pocket' is a social feed of vibe-coded mini games (Sydney Bradley - Business Insider) - Meta's new app (launched with little fanfare) is apparently focused on "interactive, playable AI generated experience[s]." Dubbed "gizmos' these aren't "games" by the strictest definition, though apparently the app can be used to make games. Ultimately, this seems like it's about merging the stickiness (i.e. player engagement) of games and game-like experiences into a social feed.

Instagram to Experiment With Longform, Episodic Series, Live Programming on TV (Lucas Manfredi - The Wrap) - This makes a certain amount of sense. If growth is slowing on mobile (and it has to eventually) then the living room is still where a minority percentage (Gen-X and younger) watch video. And with both attentive parenting and government regulations limiting teenage users of the app, what you have left are cars or the living room.

My Wife Deleted Roblox (Michail Katkoff - Deconstructor of Fun) - Roblox is a controversial platform. They openly serve (target) players who are kids. On the one hand, they offer kids the opportunities to create their own games in an open marketplace. On the other, they are bringing extractive economics to the children's media space. And they have not always had the best safety features for those kids. It feels a bit like the argument in this piece is that they are the least bad option out there. But (parents especially) read it for yourself.

CakeworldAI (Product Hunt) - this educational experience popped up in one of the AI newsletters I received. I haven’t played it myself, but I thought I'd share it here just as a potential inspiration for those interested in making educational products that could be "kid safe."

General Intuition’s $2.3B bet that video games can train AI agents for the real world (Rebecca Bellan - TechCrunch) - There's an interesting use of video game data here (though I wonder how many players know the ToS allow their controller data to be collected). Fortnite is also considered one of the proto-metaverse platforms (along with Minecraft, Roblox and others), so you've got a convergence of emerging technologies here as well.

Even Google recommends DuckDuckGo (DuckDuckGo via X)- And finally, is it a gaffe, or proof that Google maybe still isn't (always) evil? Of course, we should note that DuckDuckGo has an AI bot too, but not integrated into search the way Google is.