For journalists, the rise of AI is creating new challenges, particularly around trust with readers, viewers and listeners.

You can’t get very far into a conversation with journalists and journalism educators before artificial intelligence starts to dominate the discussion. I found myself in that conversation over dinner recently, and I asked the working editors at the table a question I’m hearing more and more as a journalism educator: What do you want the next generation of journalists to understand about how to ethically use AI?

The question isn’t an easy one to answer. Most newsrooms are still trying to figure it out for themselves. Editors are grappling with how to write policies that make room for rapidly changing technology while protecting the integrity of journalism. They know they need to get it right.

That matters because news organizations are already struggling to maintain the trust of audiences that have grown increasingly skeptical of journalism. They can’t afford to give people another reason to question whether they can trust what they read.

While editors said they were still developing their policies, some basic principles shouldn’t be up for debate. One editor put it plainly: newsrooms are responsible for everything they publish, and they can’t outsource that responsibility to a machine. That’s a line that shouldn’t be crossed. When an irate reader calls or emails to complain, can a newsroom really respond, “It’s not our fault. The AI did it”?

AI is a powerful tool, and it has a legitimate role in journalism. It can assist with tasks such as translation, transcription, research, and helping journalists understand or analyze data. But the responsibility for journalism ultimately remains with the journalist, and with the newsroom that publishes the work.

While AI can be a valuable research and reporting aid, a human journalist should be responsible for writing, editing, and fact-checking news stories. The editors also said the same principle should apply to visual journalism. AI should not be used to create images in a way that suggests they are original work produced by a human. If an AI-generated image is used in a newsletter or other journalistic publication, its AI authorship should be clearly and prominently labeled.

The challenge is that AI is advancing faster than newsroom policies can keep pace. Editors are being forced to wrestle with a fundamental question: how, where, and when does AI belong in journalism without blurring who is ultimately accountable for the work?

There are some news agencies that are embracing AI as a reporting and writing tool, and that’s where the lines are being drawn. My old newspaper, The Fresno Bee had a story with this byline: “Edited by Bryant-Jon Anteola. Story produced with AI assistance.” It was a roundup of top stories on The Bee’s site.

At the end, of the roundup was this note: “This report was produced with the assistance of a proprietary tool powered by artificial intelligence based on our own originally reported, written and published content. Before publishing, journalists reviewed this content in compliance with McClatchy Media’s AI policy.” You can read that policy here.

After half a century in the news business, I’m troubled by outsourcing news functions to a machine. I still take an old-school approach to journalism: The news should be reported, written and edited by reporters and editors. I’m also uncomfortable with an AI bot being part of the byline.

When I was coming up in the business, a byline was something you strived for. It had to be earned, and it had value for readers. They knew you could contact the reporter to praise a story under the byline or complain about it. Now, it is essentially a note to readers that the reporter used a machine to help get a story written.

Over the years, I had many byline-worthy stories that went without one because, in those early days, the editor decided, sometimes on a whim, whether a byline was going to be attached. Now, the first thing reporters do is write their byline and then the story. Yes, journalism has changed.

That doesn’t mean I’m resistant to change. Quite the opposite. I’ve watched technology transform the business in ways that would have been almost impossible to imagine when I started more than a half-century ago.

My first stories were written on a Royal typewriter. Then we transitioned to the IBM Selectric typewriter, which produced digital tape for the typesetters. From there, the business changed rapidly. Computers took over the newsroom, and the internet became the primary way we delivered the news, replacing the newspapers that once landed on your front porch courtesy of a news carrier.

The technology has changed dramatically. The fundamental responsibility of journalism has not.

AI is certainly going to further change the news business, and there are legitimate ways it can help reporters and editors, such as researching information, analyzing data, finding patterns, handling routine tasks and perhaps freeing journalists to spend more time reporting and talking to people.

But there is something fundamentally different about having a human take responsibility for a story. Journalism isn’t simply the assembly of facts into sentences. It requires judgment, curiosity, skepticism, context and, sometimes, the instinct to recognize that something doesn’t quite add up. It means deciding what questions to ask, whom to trust, what matters and what deserves further digging. Most importantly, it means someone is accountable for what appears under a byline.

That’s why I’m willing to adapt to AI, learn from it and use it where it makes sense. But I’m not ready to have AI in the byline. A byline should still mean that a reporter or editor stands behind the work—and is willing to put their name and reputation on the line for it.

I’ve been looking for newsrooms with solid AI policies. I like what the Associated Press has set out for their journalists:

“AP today released updated newsroom standards governing the use of artificial intelligence, reinforcing that AI can assist journalists with specific tasks. Editorial judgment, verification and accountability remain the responsibility of AP journalists.”

Among the approved uses of AI are:

  • Assisting with early-stage research and document summarization.
  • Supporting transcription and translation.
  • Suggesting headlines, story summaries and shotlists.
  • Assisting with grammar, spelling and search optimization.

“In every case, AI-generated output is reviewed and edited by AP journalists before publication. AI does not replace reporting, sourcing, editorial judgment or verification. AP continues to prohibit the use of generative AI to create, alter or enhance news photography.”

This seems like a sound approach to a tricky issue. I’d love to see the AI policies of other newsrooms. We can compile then for general use if you send them along.

(The column has been updated.)