Scott Converse has spent years launching and experimenting with local news projects, driven in part by his dissatisfaction with the state of local journalism. The longtime local media and technology figure’s latest project, Longmont News Network, uses artificial intelligence to generate stories about civic affairs in Longmont.

Since increasing its publishing frequency earlier this year, the platform has drawn attention for both its ambitions and its mistakes, including articles containing fabricated information, misspelled names, and AI-generated images that some residents mistook for real photographs.

Robin Burke, a professor of information technology at the University of Colorado Boulder, draws a distinction between news and what he calls “news-shaped objects.” AI-generated articles, he says, fall squarely in the latter category.

But Converse sees AI as the future of journalism.

In 2017, Converse founded the Longmont Observer, which later became the Longmont Leader. He also co-founded Longmont Public Media in 2020. He said he is not a trained reporter but has decades of experience working in media and technology. He previously held high-tech executive roles at numerous Fortune 500 companies, including Apple Inc. and Paramount Global, and he currently serves on the board of Longmont Public Media. In 2013, he founded TinkerMill, the largest nonprofit makerspace in the Rocky Mountain region, according to its website.

Scott Converse (Courtesy photo)

Longmont News Network maintains active Facebook and Reddit pages. While the Facebook page has existed for years, regular posting did not pick up until early January, when posting frequency increased from every few days to near-daily coverage.

When first contacted by the Times-Call for an interview, Converse requested an agenda for the call, writing in a Facebook message that “you are, technically, the competition.”

During the interview, Converse said the Times-Call played a role in his decision to launch LNN. He said he started the Longmont Observer because he was dissatisfied with the Times-Call’s local coverage and unhappy that the newspaper’s office had moved out of Longmont. The Times-Call, part of Prairie Mountain Media Publishing, is again operating out of Longmont. The local team moved into their office on South Hover Street a little over a year ago.

He said he was convinced that local news could no longer survive under its traditional business model.

AI reporters, human editor

Converse said LNN is an experiment built around what he describes as AI “agents” that act as reporters, scanning public documents, meeting transcripts, budgets and records to generate stories. He said the system includes what he calls a “dark news desk,” which searches for patterns and signals in public data that could point to future developments, such as housing projects or administrative changes, but does not publish information without human review.

According to Burke, summarizing public documents or meeting agendas misses much of what journalism is meant to do. “The fact that something wasn’t discussed is as important as what was discussed,” Burke said. “There’s a narrative about what’s happening in the city.”

Framing journalism as something that can be replicated simply by processing text, Burke said, reflects “a very impoverished idea of what journalism is.”

Converse has also circulated a 13-page business plan for a broader venture called SmartNewsNetwork, which frames LNN as a pilot project, with Longmont being the first city in what he hopes will be a nationwide phenomenon. The plan proposes replacing traditional newsrooms and reporters throughout the country with an AI-driven system overseen by a single human editor, one editor per city.

His proposal states it will seek funding primarily through the sale of structured civic data to paying clients such as developers, attorneys or governments, rather than relying on advertising revenue, subscription fees or grants, as traditional and nonprofit newsrooms do.

While different newsrooms have different policies around AI usage, staff reporters at the Times-Call and Prairie Mountain Media do not use large language models or generative AI in the writing or reporting process. Reporters may use copy editing programs for spelling and grammar checks that have an AI component, or an AI-assisted transcription program for transcribing interviews automatically.

Converse’s business plan repeatedly describes Longmont as a “news desert,” a term typically reserved for counties without any daily or weekly newspapers. According to an interactive tool on the University of North Carolina’s Hussman School of Journalism and Media, which tracks news deserts nationally, Boulder County would not meet that definition because it has numerous daily and weekly newspapers, including the Times-Call and its sister publications the Daily Camera and Colorado Hometown Weekly; Yellow Scene Magazine; and a recently opened Axios bureau. Longmont also has the Longmont Leader and Longmont Public Media, which are not print newspapers but still provide coverage.

Burke said claims that AI-generated platforms can solve journalism’s financial crisis often rest on what he described as a shallow understanding of how reporting actually works. “Where do the facts come from?” he said, arguing that someone still has to be present to observe events, verify information and decide what matters.

Without that labor, Burke said, AI systems risk relying on the work of existing journalists rather than replacing it. “If what you’re doing is taking somebody else’s journalistic output, then you’re not actually solving any news desert problem,” he said. “You’re creating one by being a parasite on top of somebody else’s labor.”

AI hallucinations

LNN has drawn attention not just for its business model but for errors that appear to stem from AI hallucinations.

In a Jan. 15 article shared on Facebook, LNN used an AI-altered image of Longmont’s City Council chambers that appeared to depict a renovated room. A commenter wrote that “Several (people) asked about it and even complained about money spent to change a relatively new chamber,” suggesting readers believed the image was real and that the city had used funds for renovations. Converse later removed the image and replaced it with an AI-generated infographic.

That same article misspelled Council member Alex Kalkhofer’s name, misspelled Assistant City Manager Sandi Seader’s name, and referred to a nonexistent “Council Member Mitzi Marcen.” The article also ended with what appeared to be an unfinished AI prompt: “Would you like me to draft a summary of the specific federal grants the council discussed monitoring during the retreat?”

On Reddit, LNN has faced similar criticism. In a Feb. 1 post about Flock safety cameras, one commenter with the username EagleFalconn wrote that the article was more like an opinion piece than straightforward reporting and said, “I don’t really care what an AI recommends in this context. I’m much more interested in news — a straightforward recounting of the facts.” The same Reddit user, responding to a post by LNN requesting reader feedback, wrote they were “wary of how much of this stuff is just conspiratorial hallucination.”

In a Jan. 20 article titled “A winter without snow: How Longmont is feeling the weight of Colorado’s 2026 snow drought,” an entire section referenced drought concerns that were not raised during the City Council open forum it purported to summarize. A Reddit user who said they listened to the full meeting wrote, “There was not a single mention of water. How much of this is hallucination?” Converse replied in the thread, apologizing for the error and acknowledging it was an AI hallucination.

During his interview with the Times-Call, Converse said the errors were early bugs and that he is confident that LNN now has a 99% accuracy rate, citing new safeguards built into the system. He did not provide independent verification of that figure.

Converse also questioned whether he would continue the project long-term, citing his age and uncertainty about whether the effort is worth sustaining.

“I don’t know if I want to do it yet,” he said, explaining that he is mostly retired. “I don’t need the money. I don’t want a job.”

The AI-generated newspaper Scott Converse sent to the Times-Call on Wednesday.

Less than 30 minutes after the interview ended, Converse emailed the Times-Call an AI-generated article summarizing the conversation, along with an AI-generated mock newspaper front page. The article stated, “The interview reveals a fundamental tension between the technological optimism of an engineer and the ethical skepticism of a traditional reporter.”

In the article, different sections were assigned a “confidence level indicator” score from 0 to 1 based on the LLM’s confidence in its own accuracy. A score above 0.8 means it’s “printable,” Converse wrote in his email, suggesting that AI was doing its own fact-checking. Despite the process taking under 10 minutes, according to Converse, the AI assigned a confidence score over 0.8 to nearly every section.

While some LNN articles cover topics similar to those reported by the Times-Call, the Times-Call found no evidence of plagiarism. Converse said his AI agents generate all content, which he then fact-checks for accuracy before publishing. Some of his articles cite other outlets, including Yellow Scene Magazine and the Longmont Leader, as sources.

Although Converse is quite open about LNN containing AI-generated content, it is not always clear whether readers understand that distinction. When LNN updated its profile photo recently to an AI-generated image of a reporter holding a microphone, one commenter asked, “Is this an AI picture, or is this really you, whomever you are?” (Converse has remained anonymous until now.)

‘Looks like journalism’

Concerns about AI-generated journalism extend beyond Longmont. Casey Fiesler, an associate professor of information science and computer science at the University of Colorado Boulder who studies technology ethics, said presenting AI-generated content in the format of a newspaper is troubling.

“If you see something that looks like journalism, then you have an expectation that the information is going to be accurate,” Fiesler said. She described the approach as “highly irresponsible” without plenty of clear warnings about AI limitations, adding that not all readers understand how often AI systems can be wrong.

The AI-generated profile photo on Longmont News Network’s Facebook page.

Fiesler said she is particularly concerned about a future where journalists become primarily fact-checkers for automated systems rather than reporters who gather information and engage with communities.

“There are some things that I think people want humans to do,” she said. “I think that a lot of people want their news written by humans.”

At the end of the interview, Converse offered a long reflection on his motivations.

“I don’t think there’s a story here,” he said. “I really believed the internet was a good thing. I really did. I really believed it was going to help people.”

He described newsroom decisions made in the mid-1990s to give content away for free and rely on advertising as a mistake, adding that it contributed to his decision to go on the record despite his initial hesitation. “My generation f—ed the world up, and I am so sorry.”

He said Longmont News Network is an attempt to help fix what he believes to be broken, and to atone for “evil things” he did as a senior executive for numerous large corporations.

According to Fiesler, the stakes are higher than a single website. “Our information ecosystem right now is a total mess,” she said, adding that journalism relies on trust built through human judgment and accountability.

As AI tools move deeper into news production, she said, the challenge will be ensuring that speed and efficiency do not come at the expense of accuracy and public understanding.

© 2026 the Daily Times-Call (Longmont, Colo.). Visit www.timescall.com. Distributed by Tribune Content Agency, LLC.