CNN
—
Local news sites have been popping up all over the country, and their stories seem to cover the kinds of topics any local media would care about: crime, local politics, weather, events, etc. “In-depth coverage of your neighborhood,” the media's slogan proudly proclaims.
But if you look closely at bylines by Sarah Kim, Jake Rodriguez, Mitch M. Rosenthal and others on local sites and national networks, you'll see little badges with the word “AI” on them. These are not real bylines. In fact, the names don't belong to real people. The articles were written using artificial intelligence.
Hoodline is not the first or only news site to use AI, and news organizations around the world are grappling with how to use the rapidly evolving technology while also not being overwhelmed by it.
But experts warn that over-reliance on AI could undermine a news organization's credibility and accelerate the spread of misinformation if not tightly managed. Media companies that have integrated AI into their news publishing have also seen it backfire and result in public embarrassment. An AI-generated article by technology media outlet CNET led to factual errors. Gannett, owner of the largest newspaper chain in the U.S., halted an AI experiment covering high school sports games after public ridicule. Sports Illustrated removed several articles from its website after it was discovered they had been published under false author names.
Hoodline was founded in 2014 as a San Francisco-based hyperlocal news outlet with a mission to “cover the news deserts no one is covering” and once employed a newsroom full of human reporters. The company says it has since expanded into a national network of local websites covering news and events in major cities across the country and reaching millions of readers each month.
But last year, Hoodline began filling its site with AI-generated articles, with a disclaimer page linked at the bottom of the page warning readers that “while AI may be helping in the background, the essence of our journalism is driven by real human insight and judgment, from conception to publication.”
Zachary Chen, CEO of Hoodline's parent company Impress3, which acquired the site in 2020, defended the site's use of AI and its transparency to readers, telling CNN that the media is providing valuable reporting in news deserts across the country and generating revenue to hire more human journalists in the future.
Hoodline has “dozens of editors and dozens of journalist-researchers working full time” on staff, Chen said, adding that the publication also has “a growing number of on-the-ground journalists who research and write original stories about local locations,” pointing to recent stories about restaurants, retailers and events in the San Francisco area.
Hoodline
If you take a screenshot from Hoodline's website, you'll see the article bylined “AI.”
But until recently, the site has further blurred the line between reality and fantasy: Screenshots taken last year by the Internet Archive and local outlet The Gazetteer show that Hoodline has further embellished its AI author's bylines with supposedly AI-generated likenesses of real people and fake biographical information.
“Nina is a veteran Bay Area writer who writes about great food and drink, fascinating technology, and exciting business,” one biography reads.
The fake mugshots and bios have since been removed from the site, and machine-assisted articles now have a small “AI” badge next to their bylines, but the human names remain. Archived screenshots have also been removed from much of the internet. Mark Graham, director of the Wayback Machine, told CNN that the archived pages of Hoodline's AI writers were removed last month “at the request of the site's rights holders.”
Cheng acknowledged that the company had requested that the site's archived screenshots be removed from the internet, saying, “Some websites are using old screenshots from months or even years ago to mislead people about our current practices.”
But experts have expressed alarm at Hoodline's approach, warning that it illustrates the potential pitfalls and dangers of using AI in journalism and could undermine public trust in the news.
Peter Adams, senior vice president of the News Literacy Project, which aims to educate the public on how to spot trustworthy information, said the site's use of AI and disclosures intentionally misleads readers by “mimicking” the look and feel of “a standards-based local news organization with real journalists.”
“This is a very opaque way of giving people the impression that they are reading actual reporting by real journalists who care about fairness, accuracy and transparency,” Adams told CNN.
The little “AI” badges that appear next to fake author personas on the site are “an empty gesture toward transparency rather than actual transparency in action,” Adams added.
Cheng declined to say what AI system Hoodline employs, saying only that “we combine our own custom-made software with cutting-edge AI partners to create publishable, fact-based articles.” Each article is overseen by an editor before publication, Cheng said.
The Gazetteer previously reported that at least two Hoodline employees said on LinkedIn that they were based in the Philippines, far from the U.S. cities covered by the outlet. Chen did not respond to CNN's questions about the staff or their locations.
The News/Media Alliance, which represents more than 2,200 US publishers, supports news organizations in taking legal action against AI developers who scrape news content without permission. Daniel Coffey, the group's chief executive, told CNN that Hoodline's content “likely violates copyright law.”
“This is another example of stealing our content without permission or compensation and then competing with the original article,” Coffey said. “Without quality news to begin with, this type of content, like any other practice, becomes unsustainable over time and quality news disappears.”
Chen told CNN that he takes copyright law very seriously and that Hoodline has “put in place strict guardrails and significantly improved our processes.” He asserted that the site's readers “appreciate the impartiality of our AI-assisted news,” and claimed that Hoodline's visitor numbers have soared 20-fold since the company was acquired. (Chen declined to provide traffic figures.)
That doesn't mean AI has no place in newsrooms: It can help journalists with research and data processing, reducing costs for an industry struggling with budget cuts. Some news organizations, such as News Corp., are increasingly forming lucrative partnerships with AI developers such as OpenAI to bolster the knowledge base of their large-scale language models.
But Felix Simon, a research fellow in AI and digital news at the Reuters Institute for the Study of Journalism at the University of Oxford, said Hoodline's use of machine-written articles under human names was not the right approach.
“Using AI to save local journalists time so they can focus on more thorough investigations is qualitatively different from churning out a huge number of low-quality stories that don't do anything to provide people with timely, relevant information about what's happening in their community or to help them better understand how what's happening around them affects them,” Simon told CNN.
A survey conducted by Simon Toff and Benjamin Toff, a journalism professor at the University of Minnesota, also found that the public is not accepting of the use of AI in news reporting.
“We found that people's trust in news labeled as AI dropped off somewhat, and there are reasons to think that people will not be willing to pay for purely AI-generated news,” he said.
It's hard to find an article in Hoodline's network of local news sites that hasn't been written with the software. Much of the site's content appears to be direct rewrites of press releases, social media posts, or material gleaned from other news organizations. Chen said the outlet aims to “always ensure proper attribution” and follow “fair use” practices.
“Local news has been in a severe decline for the last 20 years, but as Hoodline has grown, we've been able to bring local news to so-called 'news deserts' that gives insight into what's going on at a hyper-local level,” Chen said.
Chen said the profitable publication plans to evolve its current AI persona into an “AI news anchor who delivers news through short videos” and hire more human journalists. The plan, he said, will involve the use of fake bylines on the site, which will eventually turn them into AI news readers.
“It wouldn't make sense for an AI news anchor to be called 'Hoodline San Francisco' or 'Person A researches and Person B edits,' which is what we're aiming for,” Chen said.
In a recent column for the San Francisco Chronicle, former Hoodline reporter Nuala Bishali wrote that it's “surreal” to see her old job being replaced by AI.
“Old-fashioned, on-the-ground journalism has been replaced by fake people who have never set foot in the areas they cover because they don't have feet,” Bishali writes.
But Hoodline's changes show that a bigger solution is needed to preserve vital local news coverage.
“Without significant change, journalism as we know it will continue to decline,” she wrote.
“And it's not just small media outlets like Hoodline that are at risk of being wiped out or turned into zombies by AI.”
