A Journalist's Guide To AI

What happens when AI tools move faster than the rules meant to govern them, and what does that mean for journalists?

Yumi Wilson talks with Nat Rubio-Licht, senior reporter at The Deep View, about AI safety, regulation, agents, misinformation and the growing pressure artificial intelligence is putting on news organizations.

Nat covers AI through the lenses of ethics, safety and accountability. She explains why journalists need to understand not only what these tools can do, but also what happens when AI agents gain access to data, systems and permissions, and why human judgment remains essential.

In this episode:

How AI agents can help and create new risks

Why regulation and oversight are struggling to keep pace

Permissions, data access and agent security

Zero-click search and what it means for publishers

Responsible uses of AI for reporting, research and transcription

Why journalists must verify AI-generated claims and sources
The takeaway:

AI can help journalists work faster, but it cannot take responsibility for the work. Journalists still have to check sources, challenge outputs, protect sensitive information and make the final judgment.

About Nat Rubio-Licht

Nat Rubio-Licht is a senior reporter at The Deep View, covering artificial intelligence, enterprise technology, ethics, safety and the real-world impact of AI.

Learn more
The Deep View:

https://www.thedeepview.com/
Read Nat’s article, “The race for power behind AI’s utopian story”:

https://www.thedeepview.com/articles/the-race-for-power-behind-ai-s-utopian-story

About the show

A Journalist’s Guide to AI explores how artificial intelligence is changing journalism while keeping human judgment, ethics and trust at the center. Follow the show wherever you listen to podcasts.



Follow Yumi:
https://yumiwilson.me
https://www.linkedin.com/in/yumiwilson/

Sponsored By: Saspod

https://saspod.com/register?refcode=ProfYumiQcLU

In this episode of A Journalist’s Guide to AI, Yumi Wilson talks with communications strategist and Unspun AI founder Nikitha Lokareddy about what happens when communicators move from using AI to building with it.

Nikitha shares how she built Unspun AI, an AI-powered news intelligence tool that delivers daily briefings, compares sources, analyzes media bias and tracks social signals. She also explains why subject-matter expertise matters, how AI summaries are changing search and why journalists, students and communications teams need to understand SEO, AEO, GEO and zero-click discovery.

This conversation is for journalists, educators, communicators and organizations trying to use AI responsibly without losing trust, voice or editorial judgment.

Topics include:
Building with AI without a formal coding background
Unspun AI and news intelligence
AI summaries and zero-click search
SEO, AEO and GEO
Source comparison and media bias analysis
AI fluency for students and early-career communicators
Why human judgment still matters
Learn more about Nikitha Lokareddy:
https://www.nikithalokareddy.com/

Check out Unspun AI:

https://unspun-ai.com

Learn more about Yumi Wilson’s AI workshops, webinars and consulting:

https://www.yumiwilson.me



Follow Yumi:
https://yumiwilson.me
https://www.linkedin.com/in/yumiwilson/

Sponsored By: Saspod

https://saspod.com/register?refcode=ProfYumiQcLU

What happens to a journalist's interview after it is uploaded to an AI transcription tool?

That question goes straight to the heart of one of journalism's oldest promises: protect your sources.

In this episode, host Yumi Wilson sits down with Good Tape CEO and co-founder Lasse Finderup and puts to him the questions every journalist should ask any AI transcription company, including his own:
Who owns the data?

Can the company train on it?

Where does the recording go, and does it leave the jurisdiction?

Who can access it?

And how do we protect sources while still using tools that speed up reporting?

Lasse's most practical advice: before you trust any tool with a confidential interview, read its terms of service and search for three words: “train,” “rights,” and “perpetuity.” What you find (or don't) tells you whether your recordings stay yours. Source data, he argues, deserves far more care than ordinary workplace files, because you're handling it on someone else's behalf.

The conversation also explores Good Tape's newsroom roots at Zetland in Copenhagen, where it was built; its EU-based, GDPR-compliant approach to privacy; why Lasse refuses to advertise a tidy accuracy percentage; audit trails that let you trace an AI-generated summary back to the exact moment in the recording; multilingual transcription; and a new feature in beta, working name "artificial memory,” that connects your past interviews to future reporting.

One example Lasse offers: interview a source today, interview them again a year later, and the tool flags that they've contradicted what they told you before. That contradiction is a story in itself.

This episode marks a new chapter for A Journalist's Guide to AI : a new production team, a renewed focus, and the show's first sponsor, Good Tape, a company built by journalists, and one Yumi holds to the same standard she urges you to apply to everyone else.

If you use AI tools for interviews, transcription, research or reporting, this conversation is for you.

Guest: Lasse Finderup, CEO and Co-Founder of Good Tape
Host: Yumi Wilson, Broadcast Journalism Professor, San Francisco State University
Podcast: A Journalist's Guide to AI

Please subscribe for more conversations about how AI is changing journalism, storytelling, trust and the future of media. To learn more, check out: yumiwilson.me or connect with me on LinkedIn!

A note on timing: This episode was recorded in March 2026. During the conversation, we touched on Anthropic's decision to turn down a Pentagon contract over the Pentagon's redlines on surveillance and autonomous weapons, and OpenAI's decision to accept a similar deal. That reporting was accurate at the time, but the situation has continued to evolve since then, including renewed negotiations and additional government AI contracts. For the current state of play, consult up-to-date reporting.

*These show notes were drafted with AI assistance and reviewed, revised and approved by Yumi Wilson.




Follow Yumi:
https://yumiwilson.me
https://www.linkedin.com/in/yumiwilson/

Sponsored By: Saspod

https://saspod.com/register?refcode=ProfYumiQcLU

What happens when we start relying on AI a little too much?

In this episode of A Journalist’s Guide to AI, I talk with Max Spero, co-founder of Pangram Labs, about AI verification—and the deeper questions behind it. Not just can we detect AI-generated text, but why it matters for how we learn, communicate, and trust one another.

Max describes AI as “a fairly new paradigm,” and one of the risks he keeps coming back to is subtle yet serious: when we overuse AI, we may not learn as effectively as we think.

There are millions of AI use cases and no shortage of people building new tools. But Max says the reason he started Pangram is a bigger concern about the future of communication itself.

Some people imagine a future where AI agents mostly talk to other AI agents. “To me,” Max says, “that’s a grim future.”

We talk about why many AI detection tools fail, how false positives can cause real harm in newsrooms and classrooms, and why verification should support accountability—not punishment.

This conversation is for journalists, educators, and anyone trying to think clearly about what it means to keep humans, learning, and trust at the center of an AI-driven world.

Learn more

Max Spero: https://www.linkedin.com/in/maxspero

Pangram Labs: https://pangramlabs.com

Pangram AI Detection Tool: https://pangramlabs.com/product




Follow Yumi:
https://yumiwilson.me
https://www.linkedin.com/in/yumiwilson/

Sponsored By: Saspod

https://saspod.com/register?refcode=ProfYumiQcLU

In this episode of "A Journalist's Guide To AI," host Yumi Wilson sits down with podcast producer and founder of Saspod, Bogdan Bratis. They dive into Bogdan Bratis's unique journey from his early music career to building Saspod, a premier podcast production company based in the UK. The discussion covers valuable tips for aspiring podcasters, including the importance of podcast SEO, show naming strategies, and investing in sound quality. The conversation also explores the growing influence of AI in media—how tools like ChatGPT and audio enhancers are shaping the podcasting landscape, the risks of AI-generated misinformation, and practical ways to manage your online reputation in the age of AI. Both Bogdan Bratis and Yumi Wilson reflect on the importance of authenticity, quality content, and the challenges and excitement of podcasting in today’s rapidly evolving digital world.

(00:30) Introduction & Background - Bogdan Bratis
(02:50) Transition from Music to Podcast Production
(06:53) The Importance of Curiosity and Self-driven Learning
(08:26) Podcast Discovery, SEO & Naming Tips
(13:15) AI Tools for Podcasting & Measuring Digital Presence (16:19) Podcasting vs Traditional Media: Trends & Growth
(23:23) Keys to Podcast Success & Avoiding “Podfade”
(25:58) The Critical Role of Audio Quality
(40:47) AI in Editing: Tools and Pitfalls
(49:33) Deepfakes, Audio Cloning, and the Risks Ahead
(00:17) Controlling Your Personal Brand Online (ChatGPT, LinkedIn, Wikipedia)
(55:18) The Future of AI, Media, and Podcasting
(58:22) Practical Tips: Prompting AI and Maintaining Human Voice Follow Bogdan Bratis:

LinkedIn: https://www.linkedin.com/in/bratis/
Website: https://saspod.com/podcast-production





Follow Yumi:
https://yumiwilson.me
https://www.linkedin.com/in/yumiwilson/

Sponsored By: Saspod

https://saspod.com/register?refcode=ProfYumiQcLU