What Newsrooms are Really Asking about AI
Thoughts from a peer to peer conversation
PMJA hosted an Idea Lab July 28, 2026 as a judgment-free space for journalists to ask questions, voice concerns, and talk honestly about AI's impact on our work and our communities. What became clear: no one was alone in their uncertainty and their desire to do things the right way,
Where are lines being drawn?
When journalists started listing what they're actually using AI tools for, a pattern emerged: it's not about replacing judgment. It's about replacing tedium.
More comfortable with: Process
Formatting. Data extraction. Transcription. Workflows.
- Building workflows that pull facts and structure them
- Converting audio and video to text
- Extracting data from messy spreadsheets
- Emails, scheduling, metadata, administrative tasks
Why this feels safe: A tool handles routine tedium to make room for reporting. You decide what's true, what matters, and what gets published.
Less comfortable with: Generation
Anything that creates new content for your audience.
- AI writing interview questions
- Rewriting transcripts or interview audio
- Generating story ideas or headlines
- Creating content pitched to your audience
Why this raises concerns: Do the questions reflect what your audience actually wants to know? Are you losing your voice? Are you generating journalism or just generating content?
The line one newsroom drew plainly: Process doesn't need disclosure. Generation does.
Everyone agreed disclosing AI use to audiences is key to maintaining trust. Many felt transparency was required when the line was crossed between process and generation. Your audience doesn't need to know you used a tool to format metadata. They do need to know if a machine had a hand in a story's conception or creation.
What's really worrying newsrooms
Job displacement
Journalists have seen it before: a new technology emerges, and the calls for more efficiency, doing more with less, and reducing budgets and head counts follow. Is the "AI revolution" the next "pivot to video" — a wave of tech leaders change their products and newsrooms follow, only to be left holding the bag when they change tactics again? The fear is real and based on the all-too-recent past.
Intellectual property vs. the good of the information ecosystem
Some newsrooms have decided that generative AI use cases are a no-go because they can't control what happens to their material once it is yielded to an AI tool as free input. Then again, if you're part of the web, your work could already be in the training data of Large Language Models. You could shield your site from AI tools and web scrapers — but if we all close them off from important journalism, we're keeping valuable information out of popular circulation. The question isn't simple: How do we protect our work without starving our communities of good information sources?
Data security and privacy
Modern journalism requires security and privacy for sensitive data, interviews, and documents. Do commercial tech products meet our requirements? Can major corporations be trusted with this information? Several journalists wondered aloud whether a public media coalition could develop and standardize AI tools that met everyone's security needs — instead of each newsroom trying to navigate commercial options alone.
Environmental cost
AI training and inference consume massive amounts of power and data centers are polluting communities and draining them of water. One journalist asked plainly: How can we use these tools in a more environmentally conservative way? How do we weigh the impact of our output as a community resource against the costs of the technology? Can we build guardrails into the tools we're using to balance the scales?
How can values drive newsroom policies?
Some attendees came to the conversation with policies already in place while others were on their way. All of them were being thoughtful and intentional about aligning technology with newsroom values.
Dialogue before mandate
One newsroom had a real conversation to determine their AI policy instead of dictating it from above. They brought staff, management, and union reps into the room and asked: What are we comfortable with? Where do we draw lines? How do we protect people and our work? The policy came from that dialogue, not from hype or fear.
Prevent unsanctioned use by providing training
Any decisions about what AI tools are allowed or not allowed in the newsroom must have transparent reasoning attached. Vague, blanket bans rarely work. Instead, develop training modules that explain what a tool is, what it can do, where your newsroom is comfortable using it, and where it's off-limits.
Fundamental knowledge matters
You don't need to be an expert in AI to use it wisely. But you should have enough baseline knowledge to understand what a tool is doing and how, how it could fail, and how to check its work. If your audience is asking why you used AI on a story, can you explain it clearly? If you can't, you probably shouldn't have used it.
Double down on keeping the human work human
Human judgment is where journalism happens, and that doesn't just apply to writing stories. We need to build real-life relationships with sources and community members, too. Keeping those relationships alive between meet-ups often happens online — but that doesn't mean we can put in less effort. Sure, an AI tool could write that e-mail for you, but it could also strip the message of your voice and warmth, and thus the opportunity to nourish a relationship.