Kevin Roose and Casey Newton, the journalists behind the popular podcast Hard Fork, have left The New York Times to start Machine Gods, a tech and AI podcast that launched in the week before October 26, 2026. In an interview on The Verge’s Decoder podcast, Roose said the show is a partnership with NPR and is designed to be video-first, aimed at YouTube as much as at audio listeners. Roose is also publishing a book, The AGI Chronicles, about the companies racing to build artificial general intelligence.
What Machine Gods is
According to the Decoder conversation, Machine Gods is the next step after Hard Fork, which Roose and Newton hosted while Roose was a New York Times reporter. The hosts left the Times to run an independent company. Roose jokingly said he is its chief executive because he won a coin flip, so that someone could be named for legal purposes.
On format, Roose said the team wants the show to be video-first, meaning it will care much more about how it looks and give viewers reasons to watch, not just listen. He added that the audio audience will still get a version built for them, and that in his view a podcast can no longer succeed without a video layer.
Why YouTube
Roose described YouTube as the most important media platform in the world and the most stable one to build on, noting that serious journalism is a vanishingly small share of what appears there. Part of his goal, he said, is to find a way to make video journalism sustainable economically and creatively. He pointed to The Verge’s own video work and creators such as Joanna Stern and MKBHD as examples. He also said he is enjoying the business side, which journalists are normally taught to stay out of.
An “AI realism” audience
Roose described three audiences for AI coverage: enthusiasts, people who think it is all fake, and a growing group in the middle who see the products getting more useful and want to understand them. Machine Gods is aimed at that middle. He said Newton calls the approach “AI realism”: neither promising an AI utopia nor claiming the technology will vanish on its own, but treating it as something people should understand. He noted that walkthrough videos, such as one that built a website with Claude Code, performed well on Hard Fork’s channel even when many commenters disliked AI.
The book
The AGI Chronicles, Roose said, grew from a fear that the history of the deep learning revolution would disappear because the people involved communicate in private messaging apps and short posts rather than keeping notes. As a history buff, he wanted a contemporaneous record, and the book follows the main frontier labs and how their culture shifted as AI became a commercial product. The Verge host described it as meticulously reported.
How to follow it
Because the show is new, we have not reviewed episodes. If you want other places to follow AI developments, our guides to the best AI podcasts and newsletters and to following AI news through newsletters and feeds list established options.
What to watch next
- Whether Machine Gods builds a YouTube audience comparable to Hard Fork’s.
- How the NPR partnership shapes distribution and sponsorship.
- Reception of The AGI Chronicles among readers and the labs it covers.
- Whether other prominent journalists leave legacy outlets for independent video.
Key takeaways
- Roose and Newton left The New York Times to launch Machine Gods.
- The show is made in partnership with NPR and is video-first.
- Roose calls YouTube the most important media platform.
- The pitch is “AI realism” for a curious middle audience.
- His book The AGI Chronicles covers the frontier labs.
Frequently asked questions
What is Machine Gods?
It is a new tech and AI podcast co-hosted by Kevin Roose and Casey Newton, produced with NPR and aimed at both video and audio audiences.
Is it the same as Hard Fork?
No. Hard Fork was their earlier podcast. Machine Gods is a new independent show that Roose describes as its next generation.
What is The AGI Chronicles?
It is Roose’s book about the frontier AI labs and the race toward artificial general intelligence.
