Open Model Wars + Claire Stapleton's Dishy Google Memoir + Substack's Slop Fight

**SPEAKER_1** (0:00)
The thing about AI for business, it may not automatically fit the way your business works.
At IBM, we've seen this firsthand. But by embedding AI across HR, IT, and procurement processes, we've reduced costs by millions, slash repetitive tasks, and freed thousands of hours for strategic work. Now we're helping companies get smarter by putting AI where it actually pays off, deep in the work that moves the business. Let's create smart to business, IBM.

**Casey Newton** (0:30)
Well, Willie Nelson is getting involved in the fight against AI.

**Claire Stapleton** (0:34)
I saw this. Did you see this?

**Kevin Roose** (0:35)
Yes. He has an open letter about data centers.

**Casey Newton** (0:38)
That's right. Apparently, there is a data center that is being planned for near where he grew up in Abbott, Texas, and he does not want it to be built. He said, quote, the last thing we need is a loud water thieving, light polluting data center anywhere near our town.

**Kevin Roose** (0:56)
Bar.

**Casey Newton** (0:57)
Or bar. I think here's the issue. He wants whiskey for his men and beer for his horses, but he does not want water for his data center.
He's going to have to update that song. The thing about data centers, they're always on his mind, Kevin.

**Kevin Roose** (1:11)
Is that a Willie Nelson reference?

**SPEAKER_1** (1:12)
Obviously, that's a Willie Nelson reference.

**Kevin Roose** (1:19)
I'm Kevin Roos, a tech columnist at the New York Times.

**Casey Newton** (1:21)
I'm Casey Newton from Platformer.

**Kevin Roose** (1:23)
This is Hard Fork.

**Casey Newton** (1:24)
This week, Silicon Valley is standing up for open source AI, but is the Trump administration listening? Then, author Claire Stapleton joins us to discuss her new book about what she learned leading activism inside Google. And finally, can it stop the slop? We'll talk about Substack's new AI detector.

**Kevin Roose** (1:54)
Well, Casey, have you ever signed an open letter?

**Casey Newton** (1:56)
You know, as a journalist, we're often discouraged from participating in activism, so I usually do not.

**Kevin Roose** (2:02)
Yeah, me either. But apparently, we are the only two because there are a lot of open letters flying around San Francisco these days, and a lot of people are signing them. So today, we should talk about two of these open letters that have come out in the last week. The first one came out when Jensen Huang, the CEO of Nvidia, posted a policy letter on X, opposing what he called premature restrictions on open weight models, sometimes also called open source models, although I'm sure we will get some letters from people if we don't specify that the models are not technically open source, they are open weights, but sometimes the two terms are used interchangeably.

**Casey Newton** (2:41)
Any time we say open source on the show, we mean open weights unless we say open source for real.
That is our policy going forward and you do not have to e-mail us. Now, continue, Kevin.

**Kevin Roose** (2:51)
This letter was signed by a host of big players in Silicon Valley, Microsoft, Meta, Mistral, Hugging Face. Open AI and Google were not initially signatories of this letter, but they did sign on later.
The big notable missing signature on this letter, the company that did not sign was Anthropic. That reminds me, we should do our AI disclosures. I work for the New York Times, which is suing Open AI, Microsoft, and Perplexity.

**Casey Newton** (3:18)
My fiancee works at Anthropic.

**Kevin Roose** (3:20)
This letter from Jensen Huang and the other signatories is basically all about open models and how important they are to innovation, how they can be an important path to AI safety and security, and how we need to lead in building a future based on open weights models. Casey, I want to know what you made of this letter, but I first want to know why you think this is happening now.

**Casey Newton** (3:45)
Sure. Well, the basic reason is that there is a Saturday deadline that the Trump administration set for itself to put out a new voluntary framework for the release of new models. You may remember that earlier this summer, Anthropic released Claude Fable and had to unrelease it after the Trump administration freaked out because of its cybersecurity capabilities. Then the release of GPT 5.6 sole by OpenAI was also delayed because the government had the same concerns. This generated a lot of criticism that the Trump administration had set up this de facto licensing regime that was completely opaque, like, what do these companies actually need to do to be able to release these models? And the Trump administration said, give us until this coming Saturday, and we are going to figure something out. And in the meantime, Kevin, there were growing concerns that Chinese open source models were coming close to the capabilities of Fable and sole, and that might lead the administration to place new restrictions on open source. And for reasons that we will get into, the industry completely freaked out.