Editor’s Note: This column was written by two authors. The pronoun “we” refers exclusively to the authors’ opinions.
As much as we wish it weren’t so, artificial intelligence has hit the journalism industry.
While scouring the internet for articles on the subject, we were bombarded with various headlines praising the technology as the industry’s newest, always-available, all-knowing assistant. Though news sites originally appeared to be an oasis of critical AI sentiment, we clicked on a piece by The Media Copilot, titled “Journalism’s workforce shrinks as AI and new consumer habits reshape the industry.”
However, four paragraphs down, an “Ask The Media Copilot” AI chatbot greeted us. It suggested questions to ask, including “Why are news outlets cutting jobs now?” and “What’s BBC’s future without 2,000 jobs?” For many audiences, a healthy dose of reading and critical thought will provide them the answer to these questions. However, in our paywall-ridden, dopamine-addict digital landscape, that’s a tall expectation, enough for some to forgo direct news consumption entirely.
A study from the Reuters Institute for the Study of Journalism found that news publishers expect on-site traffic from search engines to drop by 40 percent in the next three years. Many of the media executives surveyed in the study worry that the rise of AI-assisted search engine summaries are the key cause of the decline.
Various newsrooms have signed exclusive licensing deals with individual AI developers, often with the hope that their reporting will be prioritized in AI generated responses to claw back readers through hyperlinked sources. To similar ends, some of these publishers also developed their own chatbots. The Washington Post and the Financial Times both did so, respectively signing licensing deals with ChatGPT– developer of OpenAI – building self-hosted chatbots based solely on their reporting.
Evidently, the digital landscape is making it increasingly easy to gloss over the meticulously chosen words written by human journalists.
Numerous times we’ve been reminded of this reality — from parents, professors and even peers. Aiming for a career in journalism is seemingly accompanied by a hounding of various remarks, like “AI is going to make your industry obsolete” or perhaps more frustratingly, “you need to learn how to use AI now.”
AI Can’t Think. It Shouldn’t Be A Therapist.
One should note that AI as a technical term is nearly meaningless.
AI is both used to describe the code that determines how sad your roommate’s Mii is in “Tomodachi Life,” and the surveillance algorithms that take an entire data center’s worth of processing power to run.
Most people, when referring to AI in everyday contexts, usually describe generative AI, or complex algorithms that generate text, images and other forms of digital data.
Generative AI chatbots, such as ChatGPT and Claude, run on systems called large language models — statistical algorithms that predict likely sequences of words and pixels based on innumerable swathes of training data.
When Grok gives you advice about your situationship, it doesn’t actually know the meaning of “left on read,” nor does it know how long you should wait to respond without seeming clingy. In fact, the technology is physically incapable of knowing. It’s an algorithm spitting out the combinations of words that frequently show up in the countless articles, social media posts and books fed into it.
The data sets used in this training are gargantuan, often measuring in the petabytes — equal to 1,000,000 gigabytes — for the more complex models. Data sets of websites are typically sourced from collections harvested by web crawlers, automated software that scrapes and downloads webpages from the internet.
Many of these crawlers, such as the ones used by Common Crawl, one of the largest providers of AI-training data, bypass paywalls and retain stolen material for years after publishers requested their removal. The practice is so prolific that it’s the core issue at the heart of an ongoing three-year lawsuit brought by The New York Times against OpenAI and Microsoft.
In 2023, The New York Times filed a lawsuit accusing OpenAI of illegally accessing its editorial content to train ChatGPT models and draw traffic away from its publications. Trump’s Department of Justice weighed in on the case in 2026, issuing a statement in support of OpenAI.
A spokesperson from The New York Times responded, criticizing the Trump administration for “siding with a handful of trillion-dollar AI companies at the expense of the countless American creators whose work they stole.”
This statement came almost a year and a half after The New York Times signed a deal with Amazon to license the paper’s editorial content for use on the company’s AI platforms. Amazon, alongside OpenAI and Microsoft, all develop AI capabilities for the U.S. military.
Like The New York Times, newsrooms across the nation have chosen sides in the AI bubble.
USA Today signed a deal with Perplexity in October 2025, giving the company access to its reporting and the work of 27 local newsrooms owned by the network. Their statement described the partnership as “includ[ing] the voices from local communities” and “championing local journalism.”
It made no mention of how USA Today plans to protect the jobs of its journalists, nor did they seem eager to mention Perplexity’s deal to develop an AI search engine for President Trump’s Truth Social made two months prior.
As of publication, the exact financial value of these deals are unclear, but many are reported to be in the hundreds of thousands of dollars.
What’s AI’s Place in Our Reporting?
Legacy news networks are progressively building relationships with AI tools and the controversy-ridden corporations that own them, oftentimes in the name of protecting the journalism industry. At the same time, aligning with these AI companies means aligning with the reckless behavior of Silicon Valley and the rampant violence of the Trump administration.
In lieu of these ethical concerns, legacy news networks from The New York Times to NPR have developed guidelines on how their reporters should use AI. Journalists, from skeptics to glazers, tell us that AI is here. They say we cannot simply shove our heads in the sand and avoid AI entirely. As much as we’d love a Wall Street Journal editorial titled “Fuck AI, Don’t Use It,” such a stance would likely leave our industry impotent and fall behind in the capitalist hellscape.
The current aim, then, is to create a coherent set of ethical standards regarding AI use that maintains journalistic principles and the safety of the communities we report on.
Yet, ethical guidelines can only go so far when the tech they regulate comes pre-packaged with stolen intellectual property, backsliding climate progress and Orwellian-style surveillance networks.
Thus, individual reporters are left with the decision to use AI in our reporting, and determine to what extent, if any.
On one hand, we see how the elimination of monotonous data collection or the specifics of coding in HTML for a publication’s website can be a massive weight off of a reporter working long hours with little to no pay. According to Thomas Sawano, an associate editor at California Magazine and a City on a Hill Press (CHP) alumnus, AI can act as a bolster in the absence of a larger staff.
“We’re kind of in a contracting industry, which has been the story of journalism for what seems like several decades,” he said. “Now, AI is kind of like a force multiplier, for gathering information.”
On the other, it’s a band-aid solution for an industry already hemorrhaging full-time positions, a technology actively killing our planet and public trust in the media.
In our contracting industry, young journalists are being expected to know how to take pictures, build a website and scrape government data, all on their own. AI can make that easier to digest, but it’s discouraging to know that you may need to make some ethical exceptions to have a career.
Within City on a Hill Press, both of us have felt pressure to expand our skill sets beyond the roles we applied for. We’ve picked up cameras, tried writing ledes (to varying degrees of success) and toiled away in the trenches of InDesign — oftentimes to the horror of our fellow staff who immediately told us (too kindly) how to improve next time.
It all stems from a genuine passion to immerse ourselves in news production from start to finish, but also a real anxiety that our resumes won’t be complete if we just do one thing. Marketability is a tantalizing prospect, so much so that sometimes it feels worth sacrificing our morals.
At times, we feel we’ll never learn it all, because before we do, we’ll be surpassed by another industry newcomer and their chatbot companion.
Meanwhile, shareholders that value profit over all else are eager to replace workers with robots in the name of “lower personnel costs.” Now, or so the advertisements make it seem, for the price of the Amazon rainforest, you can replace your newsroom staff with a quasi-omnipresent black box of code synthesis.
That being said, individual guilt is not a remedy to an entire industry in disarray.
Putting the onus on one journalist for utilizing a resource that makes their work manageable is not the answer. We must shift our blame and attention to the larger system forcing us to up our productivity to an unsustainable degree, and directly confront why we live in a world where sheer output is more important than any one individual’s story being told.
Because what AI is fundamentally incapable of is the reason we both come back week after week to this demanding industry.
When posed with the existential threat we believe AI poses to our industry, City on a Hill Press alum and current Berkeley School of Journalism graduate student Bomi Choe replied with something we will keep at the forefront of our minds.
“We’ll always need human connection to write human stories. That’s never going to change,” they said.
And for now, they’re right.
AI can’t reminisce with a longtime local business owner facing forced closure, and AI can’t sit down with a longtime Watsonville resident who worries about the infestation of surveillance technology in his community.
AI can’t be your one obnoxious friend that’ll stop someone midway through a sentence and ask if what they said is newsworthy, then develop it into a story. But we will be.
The truth at hand is that we live in a world that requires journalists to pitch “why they matter” in a way that’s compelling. We live in a world that aims to redefine the meaning of personhood to make us easier to separate and manipulate.
We simply cannot compete with the speed that AI processes information and “learns” to do new things. We cannot plug into a data center and scale up how fast we gather news.
What’s left, then, is our ability to be parts of a community. A community where we are able to look each other in the eye and know we’re both humans who can sit with a curiosity generative AI could never emulate.