AI is changing the newsroom, but not what makes a good story

• Sarah Needleman, senior leadership and workplace correspondent at Business Insider, which she joined last year after 23 years at The Wall Street Journal
• Peter Coy, contributor to The New York Times and The Wall Street Journal, former New York Times Opinion columnist, and author of Economics Without Numbers (W.W. Norton) due out in April 2027
• Gary McWilliams, former Houston bureau chief at Reuters and longtime editor and reporter at The Wall Street Journal and BusinessWeek, now developing business book projects

Here are five takeaways from the conversation:

1. AI is a partner in the newsroom, not a replacement

Gary sees AI as an inflection point on the scale of the internet. “At a wire service, it can scan announcements and publish a headline and summary in seconds … and handles administrative work like building the next week’s story schedule,” he said.

Peter warned about the flip side, which he called “cognitive surrender”: letting AI do the thinking while you simply type. He described building his own website with AI’s help. Sometimes it led him somewhere he couldn’t have gone alone, and sometimes he had to correct its mistakes.

“There are bad ways and good ways to use AI, and we should always be looking for the good ways, where it complements and supplements us rather than displacing us,” Peter said.

Both Business Insider and Reuters encourage reporters to use AI. Sarah uses it for research and as a sounding board. “My editor may be busy, but AI is always around.” She added: “If you’re smart about it, it could be to your advantage. If you’re sloppy, it could get you in a lot of trouble.”

2. Verification has never been more important

No one on the panel publishes anything AI surfaces without checking the original source. Peter noted that AI can be adept at inventing URLs. “Don’t just tell me what you think is the truth. Show me your sources, and then we’ll go read those sources,” he said.

Sarah described how AI recently pointed her to a Gartner study. She then called the firm to ensure she was describing it accurately. Taking a statistic straight from a chatbot, she said, “would just be the most irresponsible thing I can think of.”

Gary said the stakes for survey data have risen because falsified surveys are on the rise. Before he quotes a vendor’s research, he wants to know who was polled, how many people, when, and the method. He also checks whether a survey runs every year, so he can track how the answers have changed.

3. AI-assisted pitches are fine. Bad pitches aren’t.

The panel was unbothered by AI in pitch writing. “If the pitch is good, it’s good,” Sarah said. Peter agreed: “I don’t care that much if a PR firm uses AI in writing pitches. I just want the (accurate) information.”

That said, Sarah deletes about 99% of the pitches she receives and offered a list of what sends e-mails to the trash:

• Pitching a reporter their own story. One pitch she received opened with her own story lede, word for word, then cited her own statistics. “It’s like I’m being plagiarized and pitched my own work.”
• Commentary on a story that’s already run. She won’t rewrite a published piece to add a client’s view.
• Stale news pegs, such as a survey released a year ago.
• Unsolicited written statements. Sarah wants to talk to the actual person, especially someone she doesn’t know. Peter noted that he uses e-mailed comments only from sources that he has reached out to.
• Mass e-mails, which she can decipher if the greeting is addressed to “Sarah E.”
• Pitch length. “If I have to use an LLM to understand your pitch, that’s not ideal,” Sarah said.

4. What reporters still want: real data, exclusives and expert availability

The panel laid out what still gets a reporter’s attention in this new era.

Find the story in the data. Sarah recently built a story around a single statistic: roughly a third of surveyed employers are building AI tools in-house instead of buying them from vendors. She found the survey herself. The publisher never pitched her. “If a smart PR person looked at the data like a journalist and said, ‘I see the story in this,’ that’s when you call me up,” she said.

Offer an exclusive, with time to report it. Sarah prefers exclusives to embargoes, and the more lead time, the better. A strong data point may need a week of reporting to find examples.

Do original research on the reporter. Gary recommended using AI to scan what a reporter has written recently and find angles that other outlets haven’t covered.

Tell reporters who your experts are. A short note saying a client can speak to specific topics on a reporter’s beat, with no story attached, is welcome. Sarah files those notes and sometimes calls back months later.

5. Access and relationships matter. And make it easy for reporters to reach you.

Peter said companies have become far harder to reach. He said some of that traces to disclosure concerns under Sarbanes-Oxley, and some to a broader distrust of the media. One of his real annoyances: company websites that list only a generic e-mail address, where replies can take weeks.

Sarah described going to great lengths to get a company’s side of a story, including finding comms leaders on LinkedIn and tracking down their phone numbers. Business Insider wants to give companies a chance to comment rather than report that they didn’t respond. Peter said it’s “preposterous” that she has to work that hard to reach a company.

On building relationships, Sarah has little time for coffee meetings and won’t share what she is working on. Instead, relationships grow from a track record of relevant pitches and useful experts. Peter recommended low-pressure settings like the upcoming New York Financial Writers’ Association’s Financial Follies (Nov. 20), where reporters and communicators can talk with no story on the line – at tables hosted by companies and their communications teams.

Gary said: “It’s a relationship business, first and foremost.” He values contacts who can connect him with executives willing to share strong opinions and back them up with examples on short notice.

AI will keep changing the dynamics in newsrooms, but the fundamentals have not changed: what makes a good story, the need for verification, and the importance of access and relationships. As we explore new ways to work, let’s keep the fundamentals in place.

Watch the virtual panel

Get on-demand access to the recording here.