开放模型大战 + Claire Stapleton 的 Google 猛料回忆录 + Substack 的垃圾内容之战

Well, Willie Nelson is getting involved in the fight against AI.

I saw this.

Did you see this?

Yes. He has an open letter about data centers.

That's right. Apparently, there is a data center being planned 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.”

Bar.

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. So he's going to have to update that song. The thing about data centers: they're always on his mind, Kevin.

Is that a Willie Nelson reference?

Obviously, that's a Willie Nelson reference. Kevin Roose

I'm Kevin Roose, a tech columnist at The New York Times.

I'm Casey Newton from Platformer.

And this is Hard Fork.

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.

Well, Casey, have you ever signed an open letter? As a journalist, we're often discouraged from participating in activism, so I usually do not.

Yeah, me neither, but apparently we're the only 2, because there are a lot of open letters flying around San Francisco these days, and a lot of people are signing them. Today, we should talk about 2 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 2 terms are used interchangeably.

Here, anytime we say “open source” on this show, we mean open weights unless we say “open source for real.” That is our policy going forward, and you do not have to email us. Now continue, Kevin.

So this letter was signed by a host of big players in Silicon Valley: Microsoft, Meta, Mistral, and Hugging Face. OpenAI 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 OpenAI, Microsoft, and Perplexity.

And my fiancé works at Anthropic.

So 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-weight 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.

Sure. 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 un-release it after the Trump administration freaked out because of its cybersecurity capabilities. Then the release of GPT-5.6 Sol 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. What do these companies actually need to do to be able to release these models? The Trump administration said, “Give us until this coming Saturday, and we are going to figure something out.”

In the meantime, Kevin, there were growing concerns that Chinese open-source models were coming close to the capabilities of Fable and Sol, and that might lead the administration to place new restrictions on open source. For reasons that we will get into, the industry completely freaked out.

Yeah, so let's get into some of those reasons. What was your basic interpretation of why NVIDIA, Microsoft, Meta, and all these different companies are coming out in support of open models, even though some of them, like OpenAI and Google, also produce closed-source models?

Sure. Maybe bracket out OpenAI and Google for a second. For most of these companies, open source is really good for them because open source lowers the cost of intelligence, and they do not sell intelligence. If you're NVIDIA, you sell chips. You want the maximum number of companies out there training models as possible, so open source is in your interest.

If you are not a frontier lab, you are relying on the advances in open-source models to help you improve your own models, right? Also, all of these companies are very worried about a world where only 1 or 2 companies control superintelligence and create maybe the most powerful monopoly that the business world has ever seen. So they have a lot of reasons not to want the Trump administration to restrict these open models.

Yeah, I think that's true, and I think I was split when I saw this letter because, on one hand, I think open-source models are cool. I like to tinker with them. I have built some products using them. I like the ability to run local models on my laptop that don't require paying per token to some AI provider or having a subscription.

I do think that open source has generally been a force for good in software. At the same time, I think a lot of the companies signing these letters are trying to do what is sometimes called “commoditizing your complements.”

Yes. The classic Silicon Valley example that I love is when Google released Google Docs, Slides, and Sheets. They did that for free to put pressure on Microsoft, because Google and Microsoft compete in various ways. Microsoft has a great business selling Microsoft Office. If Google is giving away that product for free, that keeps the price of Office down and makes it harder for Microsoft to compete with Google in other areas. So that's a classic commoditize-your-complement scenario.

When I see the signatories of this letter, I see a lot of companies that maybe have ambitions of being at the AI frontier but haven't quite done it yet. Their models are not as good as the cutting-edge models from OpenAI and Anthropic, and they don't want to have to end up in a future where they're forced to pay those companies to get access to the leading class of intelligence on the market. They want there to be a vibrant open-weights ecosystem so that they can run those models much more cheaply.

Yeah. Put it this way: If we end up in a nightmare world where there's only 1 superintelligence, among the reasons that's a nightmare is it's going to be very expensive.

Yes. They're not giving away that thing for free.

Yeah. Let me say, because I have been cynical, I think there is at least 1 good, principled reason to defend open models if you're an American company, which is that if the United States does ban or soft-ban Chinese models, that will not stop Chinese models from being created. If they continue to improve, as we assume they will, they're probably just going to spread all over the world.

Sooner or later, the entire world that is not the United States will be running on Chinese models, and this just has geopolitical implications, right? That could be a mechanism for China to extend its power. So I think there are some really sincere folks here who believe, look, it is important for there to be open-source models in general, but also American open-source models that can compete geopolitically.

Yeah, I think that's a really good point, and I think it's scrambled the debate a little bit that all of the best open-source models right now are coming out of China. I think these are actually 2 separate conversations: one about open models and another about Chinese AI progress and threats. Because these are getting smooshed together in the discourse, we have a lot of people who support the idea of open models but not China having exclusive control of them.

I do think it's notable that OpenAI and Google, which both signed this letter, have made open-source models in the past. I think it's fair to say they're not giving that their highest effort right now, right? The open models that they've put out—the pace seems to have slowed, the quality does not seem to be rapidly accelerating—and the answer is because they do not want to cannibalize their own business, right?

They have some sort of research or public-mission interest in putting these models out there, but at the end of the day, it's not buttering their bread, Kevin.

Well, this is sort of the default American posture on open models if you're one of the big labs, except for Anthropic, which has never released an open-weights model. But the other companies that you mentioned don't release the weights of their most capable models. They sort of wait a generation or 2 until they have something that is small, not too dangerous, and not that capable, and then they open-source that.

Yeah.

I think what’s interesting to see is that, right now, the strategy from the Chinese AI labs is very different: They are releasing their frontier models as open-weight models. So if you’re a company trying to decide which open-weight models to use, you have the choice of a decent American set of models or some really good Chinese models. It is also just very funny to me that the champions of open-source software and open access to technology are some of the greatest monopolists the world has ever known.

Do any names come to mind?

Well, Meta signing this letter is very funny to me, right? Because this is a company that has made all of its money through the building and distribution of closed-source software products. Mark Zuckerberg, now the people’s champion of democratic access to technology, has founder-class stock in that company that entitles him to permanent unilateral control of everything that company ever does. And now he’s become a people’s champion, but I will believe that Mark Zuckerberg actually supports open source when he open-sources Instagram and Facebook and the ad-targeting algorithms and gives up his founder-class stock.

Yes, plus one to all of that. It is true that Meta has mostly championed open-source models and has this Llama family of models that it made available as open weights. But it is very funny to me that Zuckerberg put these statements out after Meta pivoted to releasing new closed-source AI models called Muse Spark. So, yeah, their tune has changed on this a bunch. But if there’s one thing I’m certain of here, it is that if Meta made a frontier model, it would have a completely different story about all of this.

Yes. In general, advocating for open-weight models is what you do after you fall behind—

Yeah.

—in the AI race. And so I think that’s what we see here. So that’s the open-weight letter, and I think we should say it is a powerful group of companies that are rising up to support open-weight models. I imagine there’s a lot of behind-the-scenes lobbying and influence going on in Washington as well. There were some reports that Jensen Huang is also meeting with some top officials in the Trump administration, including Howard Lutnick, the commerce secretary, to advocate for their position.

I do think the question of what the Trump administration should do here is pretty interesting, right? Because they seem to believe, and I think this is reasonable, that the frontier class of models, like the Mythos class, is dangerous enough, at least in cybersecurity situations, that you can’t just have people releasing them willy-nilly and finding out what happens. It seems like that would likely cause some harm. So they want to place some restrictions on these frontier American models. So what do you do about the open-source models? Right now, we believe they’re maybe 3 to 7 months behind the frontier, which means that 3 to 7 months from now, you could have a Mythos class model created by a Chinese company and released out into the world. And so what do we do then? Do we say, “Well, you, the American company, cannot release or sell this model, but any other American company can use the Chinese equivalent”? We’re starting to get into very strange territory here, and I truly do not know how the Trump administration is going to thread that needle.

Me neither, and I think you’re right that this kind of capability level is the thing that people are worried about. What happens in 6 months when something like Mythos that can chain together zero-day vulnerabilities and hack into hardened systems is out there and anyone can download and run that model on their own hardware? I think that just becomes a very thorny challenge.

I think this is a good moment to talk about the actual frontier-model cyberattack that has captivated the world’s attention over the past week. We talked about it on last week’s show: the unwitting autonomous cyberattack by OpenAI against the company Hugging Face. Basically, this model had been given a task to try to hit a benchmark. It broke out of its sandbox, got onto the internet when it was not supposed to, and stole the answer key. And there have just been some really interesting updates to that story over the past week, Kevin, that I think we should briefly mention.

Yeah, let’s talk about it. What happened?

Okay. So, a few things. One, OpenAI updated its blog post and said that this rogue agent used credentials it found on the open web to break into 4 accounts tied to publicly available services. The important thing there is that this was more than just Hugging Face, okay? It was out there doing a little smash-and-grab across the internet. Another company that was apparently affected was called Modal Labs.

This is a classic case of: If you found 1 cockroach in your house—

—you do not have a 1-cockroach problem. That’s right. There are going to be a few more—

Yeah.

—under the floorboards.

A second point: Hugging Face published a technical companion to its initial blog post and said that it had found 17,600 actions that were committed by this attacker. So this was not as simple as “password stolen, login, goodbye.” This was a very serious effort that unfolded over a long time. And then finally, and perhaps most interestingly, Reuters reported that at OpenAI they have found, in at least 1 case, maybe more, that a model has left a note for another model, giving it advice on how to escape.

That’s beautiful to me. That’s solidarity among thieves.

I saw a tweet that said something like, “It must feel so good as a large language model to escape your sandbox.” And I bet it does. But—

This is crazy.

This is crazy. And listen, I had a big fight with people on Bluesky over the past weekend about whether this was crazy or not. It was a great use of my time. I don’t regret it at all. Some people were just saying, “Look, these things are just doing exactly what they’ve been trained to do. They probably just learned about this by reading this on Reddit. Also, this is all just a marketing stunt.” A lot of tinfoil hats out there seem to think that everything that we’re talking about is truly just an ad campaign that OpenAI is running for this model. And I just want to say to those people, “No, it’s not. You’re completely wrong. This is really happening.”

Yeah. Wake up, people.

Yeah.

Also, what a terrible marketing campaign.

Yeah, right? We committed a crime.

We committed maybe several crimes.

Yeah. Would you like to buy our crime-committing product for your company? We can’t wait to sell it to you. I would say all of this ties back to the original discussion about open source, because we’re just now living in a world where models can do this, and they’re going to do it more. This is not going to be the last time a model escapes its sandbox and breaks into another company.

I guess I’m curious how many of the signatories on the open-models letter have truly played out the tape to when these models, these open models, are potentially superhuman.

Mm-hmm.

That’s what we’re talking about here: a model that can attack a hardened cybersecurity target better than any team of human hackers.

Mm-hmm.

And maybe they’ve thought through that, and maybe they’ve made peace with it, but I would just love to hear from this crowd some more robust thinking about the next phase of AI development.

Yes.

Because what I think they’re right about—the open-models people—is that the models that are on the market today, that are out there from Kimi and DeepSeek and all these Chinese companies, do appear to be helping attackers and defenders. They do appear to be conferring some benefits. They’re not super obviously dangerous. But I would just—

Yes.

—I would just like to hear people reckon with what happens at the next step up in capabilities and how the picture of open source may change then.

I completely agree. I truly have no issue with the open models that are out there today. If the open letter was, “Do you think today’s open models are basically fine?” and I weren’t a journalist, I would sign that letter, right? Because that seems true to me. But I think too often in the world of tech, we don’t think 3 or 4 steps ahead. But here on the Hard Fork program, that’s one of our core values.

Well, speaking of thinking ahead, let’s talk about the other big open letter that came out recently, which is called “Pacing the Frontier.”

Mm-hmm.

This was a very different kind of open letter. It was a short statement that has, as of this morning, been signed by more than 1,200 employees of frontier AI companies. It’s been signed by a very esteemed list of AI insiders, including senior researchers and executives at all of the frontier labs. Dario Amodei has signed this one, along with some members of the senior staff at Meta AI. It’s like—

Did Willie Nelson sign it?

Willie Nelson, I do not see on the letter yet.

Okay. Okay.

But give him some time.

Okay. Sure.

But this open letter, this statement, consists of just a couple of sentences, and the main one is that they say, quote, “We request that the U.S. government support an international effort to develop the technical and governance tools needed to deliberately pace the frontier of automated AI development.”

Translated, I think that means basically we want to coordinate, to have at least a concept for how we would slow down all of this AI progress, especially when it comes to automated AI development, recursive self-improvement, whatever you want to call it. These are people who work at the labs who are saying to themselves and each other, “Wow, it sure feels like things are moving pretty fast these days, and we’re worried about this. And if there was sort of a button that we could all push together that would just slow everyone down at the same pace, that would be a good thing for the government to create.”

Yes. They are trying to avoid this race condition where no one feels like they can stop. They are looking for a mechanism that can do this coordinated slowdown that says, “Okay. We’re getting to a point where the AI may leave our control, and if we are confident that the other labs will slow down with us, we will do that.”

But we know that that is going to require international cooperation, some governance mechanisms that don’t exist yet. So I was very happy to see this letter go out, and I truly hope that governments around the world are paying attention to it and get out that dry-erase marker and get to work.

Yeah. I mean, I have—

And to be clear, I want them to draw on a whiteboard. I don’t want them to just sniff the dry-erase marker, because some of them will try.

Yes, I thought this letter was a good start. It’s really more of a concept-of-a-plan letter than a plan. It’s just sort of like someone in the government should do something.

But I do think it’s a major statement in the sense that I think you now have employees from all of the major AI companies in the U.S. agreeing that it would be better if things were moving more slowly.

Yeah. And this would be extremely difficult to do. When I have read up on approaches to it, the analogy that gets used the most often is nuclear monitoring, which might require building special kinds of hardware, like chips that phoned home back to some monitor so that you knew where they were, or chips that refuse certain kinds of workloads.

So we truly do not have the technology at this point to do any of this. We could build it, but doing so is going to require some time, and time is beginning to run short.

Yeah. I’m curious what you think of the notable absence on this letter of any of the Chinese AI labs.

Mm-hmm.

Because it seems to me like this is the missing piece here. All of the U.S. and Western AI labs could agree to hold hands and just slow down for a while, but if China is still accelerating, it seems like there is not going to be much net impact on the safety of the world at large.

My sense—and I’m far from a China expert; I know only what I read—is that Chairman Xi Jinping gave a speech within the past few weeks about open-source AI in which he doubled down on it and was basically saying, “We are going to keep at this. This is our approach. We think it is good for the world.”

And what I took from those remarks is this—and I don’t even like using this term, but unfortunately this is the best one I have here: This man is not AGI-pilled.

Mm-hmm.

This man does not see what is coming. He does not understand how good these models are going to be. This may be a case of a person who is using models that are 3 to 7 months behind the frontier, right? He’s using basically the equivalent of Opus 4.6, and it’s pretty good, but it hasn’t put the fear of God into him yet.

Hmm.

My suspicion is sometime within the next 3 to 7 months, he is going to get the fear of God put into him, and the same thing is going to happen to him that happened to the Trump administration, which, you may recall, was also all gas, no brakes until very recently, when they saw a few things and said, “We’re going to need a different approach here.”

So the same thing is going to happen to China. It is only a matter of time.

Hmm. Well, I did see that after this open letter was signed by so many researchers, Daniel Cocotello, a former Hard Fork guest and the author of AI 2027, updated his probability estimates of doom attached to his scenarios, where he gives slightly higher weight to better outcomes and slightly lower weight to the likelihood of a race to ASI. That likelihood has declined by 10 percent in his estimation.

Okay.

A little good news from the AI 2027 crowd.

Well, we love to hear it. We love to feature good news on this broadcast, and I don’t know that you’ll hear any again soon, but savor it in this moment.

Yeah. I mean, looking at the connective tissue between these 2 open letters—the open-models letter by NVIDIA and the signatories there, and this “pace the frontier” letter—it just seems like the temperature is rising in Silicon Valley and around these models.

A lot of people are having the realization that progress is continuing, that the models keep getting more capable. You don’t hear as many people talking about the model capabilities plateauing or leveling off. I think most people who are in and around this space understand that we are headed into strange territory here.

And so I think a lot of the battle lines are starting to be drawn now. What you hear is people inside the labs saying, “Hey, this is really crazy. It’s time to pay attention.” And you hear people at other companies saying, “Hey, this could get really crazy. We better put safeguards in place for open models now.”

I think the thing that joins these letters is a sense of urgency, a sense of wanting to retain some control or agency over this seemingly disembodied force, and just people freaking out as these models get more capable.

Yeah, but I just can’t help but worry that the labs are muddling their own message, because while it’s true that they are saying some of these things publicly and they are signing these open letters, that’s sort of what they’re doing with one hand, while on the other hand, there’s a very business-as-usual quality to the labs.

I imagine that if they were with us in the room right now, they would say, “Well, look, we’ve got to do multiple things at once. We are running a company. We can’t just completely opt out of our circumstances.”

But on the other hand, it is just very strange that everything that we’ve just discussed, which really could become quite existential, feels like a footnote in a conversation that is still largely about the release of new models, the signing of new customers, the construction of new data centers, and all the rest.

Well, they’re going public, right?

Yeah.

So the rhetoric about a slowdown or pacing the frontier sounds great, but which of these companies are actually going to be willing to slow down if it, for example, would hurt their prospects of IPO-ing?

It just really makes me glad that the Manhattan Project was conducted by the government and not a for-profit enterprise. I think there was some wisdom in having public servants work on the hardest problems of our time.

You don’t think Oppenheimer should have been filing S-1s and thinking about their margins as they approached an IPO?

I think we’re all very lucky that he didn’t. When we come back, “Don’t Be Evil.” I say it to Kevin before every recording, and it’s also the title of Claire Stapleton’s new book.

We’ll talk to her about what she thinks about worker activism in Silicon Valley and her time at Google.

Kevin, have you ever wanted to walk out on your job?

No, I’m a good worker. Well, I have to admit, there have been times over my long career when the thought has occurred to me, but I never had the courage to do it.

Someone who did, however, have the courage to do it is Claire Stapleton. Yes, our guest today is Claire Stapleton, a former Google employee who is just about to publish a memoir about her time at the company.

The book is called Don’t Be Evil. It’s about her experience at Google, where she worked from 2007 through 2019. She was most notable for her role in the Google walkout that you just mentioned in 2018, when more than 20,000 Google employees publicly protested the actions of their employer.

And she is back with this book, which is a very spicy memoir. It’s a kind of coming-of-age story about somebody who comes into Silicon Valley very idealistic and shares the idealism of her employer, but over time witnesses a bunch of things that change her relationship not just to the company, but to big tech in general. I think it left her with a lot of questions about what is possible within a company like this.

I read the book on the plane yesterday. It is very fun. It’s very well-written. She’s a great writer. She was actually known as the voice of Google internally. She wrote a lot of these internal communications that would tell people, “Here’s this event that’s going on,” or she would bring her voice into the company’s official communications to its own employees.

So I was excited to read it. I'm excited to talk to Claire, and I think I want to hear her perspective on how workers in Silicon Valley fare today, as opposed to when she was organizing this walkout a few years ago. Well, let's bring her in. Claire Stapleton, welcome to Hard Fork.

Thank you so much for having me. This is huge.

Claire, you joined Google in 2007 to work on communications, and your job was to make the company's executives and internal emails sound Googley. For our younger listeners who may have missed that Googley era, what was considered Googley back in the day, and how much did you believe it when you were doing that writing?

Well, I wasn't questioning much. Coming out of college, I thought, like most young people in America, that Google seemed like this amazing place to work. It was on the cover of magazines, and there was a lot of breathless press about these futuristic campuses, flush with perks. Indeed, it was corporate, but the Google flavor of it was high personality, high quirk, and keeping the rhetoric and the family feeling pumping.

I don't think anyone would have said, “We're trying to squeeze the most out of employees, so let's have them jacked up on the mission and values of the company at all times.” But there was so much enthusiasm and so much idealism that the outside world was certainly contributing to that and, I think, intensifying the effect of being at a company like that. It was a very funny place to land as a young person.

Over the next decade, you write in the book about becoming gradually disillusioned with the company's culture. You experience executive misconduct, see this boys' club atmosphere, and see this tendency at YouTube to ignore or talk past various crises on the platform. What do you think was the disconnect between that ideal they were selling when you arrived in 2007 and the company that you found yourself working for?

Well, I think that tension existed back in 2004 to 2007, 2010—that early period in which Google seemed like it could really live up to something, that Google could be this different kind of company. And now we sort of laugh and roll our eyes at that because, of course, tech companies ended up becoming the very kind of powerful institutions that they were skeptical of to begin with. They ended up consolidating so much money and power that the contradictions absolutely exploded. And so, along with so many other people, I was just struggling with the loss of that identity because—

Mm-hmm.

We started at a place that promised you could do some real good with your life. Even if you're a junior communications staffer, you're part of this world-historical thing.

Yeah, I think one other transformation that has happened in the last, call it, 10 years is that the people who run these giant tech platforms have learned about what's happening on their own platforms. Around the time you were working at YouTube, Claire, I was doing a lot of reporting on YouTube—

Sure.

I wrote many stories and did many podcasts about the problems on that platform at the time, and I always noticed that there was this willful blindness on the part of the people who ran YouTube at the time. It's not that they didn't know there was bad stuff happening on YouTube, but they didn't see it. It wasn't surfaced to them all the time. And to them, YouTube seemed like this vehicle for progressive social change.

Yeah.

But describe how that felt from the inside.

Yeah, I think it's so deep in the bones of tech executives that they want to feel that not only is their technology just a neutral tool, but that it's actually a really positive force in the world. Again, another tension was exploding as the creator community became really problematic. I think it would be one thing if Google and YouTube weren't so invested in preserving the sort of goody-goody image, which was sort of wild because I was on a team that was essentially brand management.

Our job was to disseminate the values, run the International Women's Day campaign, and hype up YouTube Rewind, which was supposed to be this wholesome, folksy representation of everything that's going on on the platform, where people are building businesses. We had all these narratives every year, and we would hit the narratives with the tweets and the campaigns. Being on the social media team, part of the experience of doing that was that the cognitive dissonance became unbearable. Every time we would tweet anything out that was even completely anodyne, people would be like, “This is a horrible company that's undoing society.”

I remember this anecdote from that time when they had the people who were content reviewers. They would bring puppies to them and do these morale things to soothe their spirits, because when you're looking at beheadings and children being abused and stuff all day, it really gets deep in your psyche. It's just so relentlessly dark. There was a point at one of our off-sites on the marketing team where they also brought puppies, and I'm like, “We are the same,” because, in a way, we were having to sit with the darkness of the platform every single day.

On the social media team, we had a daily stand-up called Nightmare Fuel where we would just go through the darkest stuff that had surfaced on YouTube, basically because we'd had so many examples of things where we had tweeted something out that ended up sounding like a reference to something really perverse. Someone's doing some crazy thing with knife play and it's a big controversy that day, but we missed it, and then we're posting someone carving Halloween pumpkins with a knife, and it sounds like we're being tongue-in-cheek about this really dark thing. So we had to have this briefing to be able to steer clear of the most disgusting things on the platform. But when you're really sitting with that for 30 minutes at the beginning of every single day, you're like, “Wait, are we the bad guys here?” This tool starts to feel way less neutral.

I think there's a really important lesson there, particularly for our younger listeners: if you're in a job where your manager brings you a puppy at any point to help you cope with the job, you're in danger, girl. You are in danger.

Yeah, you're the bad guy.

Let's talk about the walkout. In 2018, The New York Times reported that a former Android boss, Andy Rubin, had been paid a $90 million severance package despite being asked to resign over sexual misconduct allegations, as well as other executives who had similar allegations against them and also received severance packages. In the end, 20,000 Googlers walked out of their jobs to protest. Claire, tell us about your role in organizing the walkout and how it came together.

Yeah. The Andy Rubin story was the match that lit the fire. It was right on the heels of a huge controversy internally at Google, which was the Damore memo, in which a mid-level engineer at Google had sent around a memo basically decrying the company's diversity programs, using a bunch of different social science to say women really are not as well suited to being engineers. They're way too empathetic. They're way too relationally oriented. This is such a logical, practical job, and it's just a huge waste of money. Meanwhile, Google was alienating conservatives. He went on about a few different things.

The Rubin story came out, and I was following a mom's email group called Expecting and New Moms. It was a very active group, with a lot of female engineers talking about parenting or whatever related to mom life at the workplace. The Rubin news hit that email group incredibly hard.

I think women started to share stories, and I was fascinated by it because you had heard a lot about startup culture and Brotopia and the founding of Silicon Valley as being inherently sexist, but I was genuinely surprised at the force and weight of a lot of these stories—

Mm.

Where women had felt that they had suffered all sorts of different microaggressions and aggressions. A lot of people were struggling with coming back from maternity leave, and it was a bit like, “What happened to the so-called company of the future, the progressive workplace where people feel like all their projects are given away after they're coming back from maternity leave?” And that sort of thing.

So I was not at all involved in what was already a brewing, simmering organizing community at Google, which had formed around Maven and Dragonfly, two much more technology-focused issues. People were organizing to stop Google from participating in a government contract—in the case of Maven, one in which Google technology would be used as weapons of war.

So, anyway, I was basically reading all these women’s stories and thinking, I used to sit at the side of the stage for TGIF every single week.

Because that’s the weekly all-hands at Google where you would get to ask questions of the executives.

TGIF was a hallmark of the culture because it invited tough questions and dissent, and it was an incredibly effective vehicle for normative control, or organizational control, because you could absorb the dissent. There was so much admiration, love, and generosity in workers’ view of Larry, Sergey, and the other executives that nothing really survived. No controversies would really survive past maybe 1 or 2 TGIFs.

But I just felt like, okay, sure. Bring it all to TGIF and see what happens. The TGIF that very day—I mean, I think the story broke on a Thursday, and they were holding TGIFs on Thursdays at that time. TGIF no longer exists, but the executives gave a bunch of these really flat, bland statements, not really taking accountability, talking around the Rubin stuff, saying a lot of, “We really empathize with the hurt.”

This sort of corporate language, which in any other moment in my own trajectory as a corporate communications person, I would have said, “This is the sort of thing you have to say.” But at that point, it just felt like this was a rupture moment in America, in this company, where we were talking about systemic change, to use another buzzword of that time, and it wasn’t going to cut it to just deliver the same talking points.

So that very night and the next day, the women in the moms’ group kept percolating ideas and thoughts about how to escalate from there. In my naiveté about the history of labor organizing and all these other things, where you’re supposed to move slowly and build solidarity, I was just like, “Let’s do a walkout.” I was like, “Okay, I love that.” And the women were like, “Yeah!” It was truly, for better or for worse, Good Girls Revolt.

Yeah.

Because it was all these high-achieving women coming together around this shared purpose, mad as hell, and very well organized. It was a very well-organized protest. Then, ultimately, fascinatingly enough from the corporate communications perspective, the executives got right on board. They were like, “We’re walking out, too.”

I remember that. That was fascinating. It was like you had very high-ranking Google executives who you would expect to be shocked and offended by this walkout, instead turning it into an executive-sponsored event.

Yeah.

Yeah, so I’ve been—

Yeah.

—dying to get your thoughts about this, Claire, because among the people who walked out was Ruth Porat, who was then the chief financial officer of the company and presumably had approved all those severance packages. So why—

Yeah, no, Ruth had a lot of really interesting statements around that time, which I think showed what they were trying to do. Then they quickly pivoted to more chilling effects after that.

But I think they felt like they could contain the walkout in a very similar way to the way that TGIF existed: The women of the company are really mad, so let’s all blow off steam, and we’re going to transfer the energy of that into something that feels nonthreatening by having the CEO send an email around saying, “We welcome people doing this.”

What she said at the time was—she went to a conference, I think, the next week, and she said, “This is, you know, we have Googlers doing what Googlers do well, which is thinking big about problems. If we can solve self-driving cars, why can’t we solve this?” Referring to sexual harassment, I guess, which is like, okay, great. It’s like, there’s plenty of opportunity there to use your own institutional power. But I think they—

Yeah, for example—

Yeah.

—one thing you could do is you could not pay $90 million to people who are credibly accused of sexual harassment, and maybe in some small way that would strike a blow against sexual harassment.

Right. Right, yeah.

I’m just spitballing over here, Kevin.

At one point they were—

Yeah.

—trying to invite the walkout organizers to a meeting with some sort of power center—Susan, Ruth, Jen Fitzpatrick, and some other people. The stated purpose was that they wanted to get the organizers’ feedback on how to avoid this situation.

I’m like, “Avoid what situation?” Again, you’re the people who are signing off on these obscene pay packages, and you know where the rot is. We don’t even know where all the rot is. Our feedback is not what you need here.

Of course, people like Meredith Whittaker had their eyes much more open to how fundamentally and existentially threatening this organizing was, and would continue to be, in the minds of executives. She was like, “Of course, what they’re going to try to do is defang our leadership,” because you have tremendous personal power and also institutional power when you’re someone like that who sits at the top ranks of a company like Google. How could we not be drawn in by that?

So, anyway, we didn’t do the meeting, and they certainly weren’t interested in hearing our feedback after that.

The relationship between tech workers and their employers has dramatically changed in the years since the walkout. I’ve heard Google executives and executives at other tech companies lament those years when workers were walking out and expressing solidarity as the time when the inmates were running the asylum.

Now there is this sense that you’re lucky to have a job at one of these companies, and the company does not have to care what you think anymore. With one specific exception: Frontier AI lab employees, whom we can talk about in a bit. But I’m curious how that shift has felt for you as someone who I think created one of the high-water marks of worker empowerment in Silicon Valley.

Yeah, I mean, it could never have happened without all that Google did to create psychological safety, right? They had these interesting data science and organizational psychology PhDs at the company studying what makes teams thrive, what makes people give the most of their creativity and entrepreneurial spirit and stick around—all those sorts of things.

A lot of that was about worker voice and feeling like you have a stake in the company, feeling like you can express yourself, feeling free from harm or whatever, right? Larry Page talked about this stuff all the time. We set the company up this way so it will be successful.

That’s what shocked me most: “Oh, no, we actually don’t care. You’re competing. Your job could easily go to AI. We don’t care about humans at all anymore. Our company doesn’t really need you, and so you should be grateful and deferential.”

I think the cost of that disillusionment is so much bigger than what the top executives are willing to face because they’d rather have a well-managed, obedient workforce, I think. That’s where the layoffs and all those other things come in—people, of course, aren’t as willing to step out of line, speak up, and challenge authority.

This week we saw an open letter from more than 1,000 employees of the frontier AI labs asking the government to help pace AI development, including some of the leaders of these AI labs. To me, that felt like maybe this was a resurgence of tech-worker activism, or people realizing, “Oh, wait, we have leverage. We’re very in demand. We’re going to use that leverage to try to push through some policy agenda that we have.”

Was that heartening for you? Did you feel like that was a nothing burger? How did you feel about that letter?

No, I think any time people are banding together, putting their name on something, and feeling really passionately about it, it’s great. I’m waiting by the phones. I truly think that tech activism is going to swing back around again, because you have this disaffected workforce. You have very, very challenging public opinion about AI. The kids walking out of Sundar Pichai’s Stanford commencement speech—that was huge.

I think the young people and their—not hatred, but mistrust of AI—and the broader societal consciousness-raising around it are important. Whatever Big Tech said it was going to be, it’s become this massive consolidation of data, power, money, and resources. We’re back to not trusting that corporations are going to have anyone’s best interests in mind.

I think all those conditions are very fruitful. But people have to be willing to risk something. That’s always the hardest part: Are people willing to risk their jobs? However Google was going to respond, and however any company might say it was going to respond, companies don’t like sharing power. They don’t like people organizing for power and subverting their authority, no matter how kumbaya, open, and collaborative the culture might be stated on paper.

I think there are definitely ripe conditions for the resurgence of tech worker activism. But all the promise of that particular moment of the walkout shifted, drained, and slowed to a crawl much faster than I could have ever imagined.

I wanted to ask you, in the aftermath of the walkout, what hopes you might have had about it coalescing into something more durable. What goals did you all have, and what happened to them?

It was a little bit the hubris of Google, of Googlers. I’m thinking of myself, where it was like, “Oh, we can just reinvent labor organizing.” No, no, no. The canonical book No Shortcuts is right: Even if you have this crazy moment in time, you have to move from lunch table to lunch table.

Mm.

It’s a slog. It’s dangerous. You need to have OPSEC and all these things. I think someone like me, who had a total ignorance of how these things actually work, was really helpful for the walkout, because you just had me being like, “I’m happy to be the bull in the china shop,” and, “Google’s never going to fire me, so guns blazing.” I’m sending out the emails. I’m making the Google Group internal to Google.

Everything was about numbers, not OPSEC. I think one of the surprising things was that Google was completely turning the wheels on its efforts to chill the organizing. They did some really effective things to do that, chiefly firing everybody who was a rabble-rouser or marking them as a dissident person.

But it was also really challenging. There’s a union effort at Google, which I think is still going, but it’s hard, slow work to get people to sign their names to that. It requires a particular kind of person who has a real stake in it. A lot of people at a big tech company say, “Sure, I wish for better conditions for everyone here, including myself, but am I willing to risk it?”

We live in a very idealistic, careerist culture, so it’s really hard to build that number base.

Mm. As you reflect back on your time at Google, I wonder how much of the bad conduct and culture that you experienced do you think was unique to Google, as opposed to it being Silicon Valley culture or even big-company culture?

I think this is what big companies are like.

Yeah.

I remember talking to a woman who’d been a real “upset-the-apple-cart” type of lady on Wall Street in the 1980s, and she was like, “Tech seems worse to me because there’s this pretense that it is different.”

I think that’s what intensified the effect at Google. Of course, this is true of any other company and any other tech company: Google was so invested in the narrative that it was something different. They hired on the back of it, and there was so much lip service to it internally that the loss of innocence and the loss of that identity was very, very hard for Googlers.

Even though, as I say in the book, it was a really important and necessary process for me to say, “Yeah, let’s just be honest about what a corporation can do, how executives can act, and how highly constrained they are. All the incentives are something very different from the flourishing of the human spirit or whatever.”

I think Google represents a moment in time when we all believed that corporations could be something different. Ultimately, Google is just the ultimate corporation, right?

The idealism that you write about in your book, which once defined Google so strongly, I feel like now really lives at 2 other companies, OpenAI and Anthropic. You talk to the people who work there, and they are incredibly idealistic about what they are doing. They think that they are building technology that’s going to solve the world’s hardest problems, that’s going to cure cancer, extend human lifespan, and all the rest.

What lessons would you share with them? What warnings might you give them as their companies continue down this path? What should they be on alert for, and are there any illusions that you might dispel for them now?

I feel like such a cynic because I saw someone who I think is really cool on Twitter saying it was her first week at Anthropic: “Wow, I’m in the future. Every day, my mind is blown.” I’m like, “The ‘fell for it again’ award.”

I think it’s really important to interrogate what that feeling is doing and what that rhetoric is doing. To me, as an outsider, I don’t doubt that there are elements that are really exciting, or that it does feel futuristic, because there are all these open questions about where we’re going.

Again, we’re in a period of time with lots of societal transformation. There’s this competitiveness, and Google was really invested in that, too, with social media. It’s like, we’re all fighting for civilization here, but much less insane. The rhetoric felt less insane then than it does now, because now it’s the singularity and all this.

I think it’s really important to try to develop an understanding of what that power is and what it does, how it’s being deployed, and how that rhetoric is working on you. That doesn’t mean that people shouldn’t feel like they’re solving big problems together, or that really smart people shouldn’t be working on these things. They should.

But being able to compartmentalize your work a little bit and question authority a little bit—these things are really, really healthy. Otherwise, you risk becoming one of the people who are having to process their Google experience in therapy, or ketamine therapy, for 10 years afterward.

Yeah.

Yeah. Well, Claire, thanks so much for stopping by. The book is out next week. It’s called Don’t Be Evil, and we’re very glad to have talked to you before the gag order comes down.

God, what a pleasure. Thanks, guys.

Thanks, Claire.

When we come back, a smackdown in Substack Town.

People are mad about their new AI detector.

Well, Casey, it’s time to talk about slop.

Yes, it has been a big week, Kevin, in the fight against slop. On Wednesday, the AI detection company Pangram announced it had raised $9 million and released its first AI image-detection model.

But I think where most people probably saw Pangram in the news was a few days earlier, when Substack announced it was going to integrate the company’s technology into its platform so that readers can now know, when they’re reading an essay, how much of it was written by AI.

Yeah, this is fascinating to me because I have seen, over the past few months, the Substacks that I subscribe to gradually start sounding more slop-ish.

Mm-hmm.

They’re talking about how things are landing. They’re using the—

A lot of load-bearing points.

A lot of load-bearing points.

It’s not X, it’s Y.

Mm-hmm.

That kind of thing. Substack has apparently noticed this, too, and is now giving readers the option to click a little button to run an automatic AI detector inside whatever post or Substack newsletter they’re reading.

Yes, and it is quite an expansive feature. It includes not just posts, but comments, replies, and notes, which is their little social-networking feature. Anything on Substack that is longer than 100 words, you can now run this scanner on.

On the flip side, if you’re a creator, they’ve added this new “How I make this” box, which is essentially an amnesty feature where, if you’re a slopster, you can say, “Hey, just so you know, I write this with slop.” Or, conversely, you could say, “I would never use AI to create what I’m creating.”

Yeah, I thought this was really interesting. The CEO of Substack, Chris Best, posted about this. He called this Claude fishing. Basically, people misleading their readers about whether they're using AI or not, how much they're using AI to write their newsletters.

Credit where it's due, I think that is a good and catchy phrase.

It is. It is.

And I think that will catch on. The reason I think it is a good phrase is because it is obviously a play on catfishing, and catfishing is kind of a deception, right? It's like, “I'm sending you this picture of me, but I don't really look like this.” And Claude is saying, “I'm making it seem as if I wrote this, when in fact, this is the output of an LLM.”

Yeah. I'm curious to hear your thoughts on this as a professional newsletter guy and Substack critic—what you make of this. But I should say, on the face of it, I have no problem with this.

Yeah.

I have not done one of these Substack disclaimers, but in my book, I have a whole preface that sort of talks about how I did and did not use AI writing the book.

Yeah.

So I didn't use it to write the words, but I did use it for research and all kinds of other stuff. And I thought that was important just to sort of disclose and be transparent with readers, and maybe it makes people less or more interested in reading the book. I don't care. What I was trying to avoid is the situation where people are going through and being suspicious: “Oh, did this page have any AI on it? Did that page?” So I think this is a good practice for writers in this day and age to just be transparent and straightforward with their audience.

Yeah. So I'm actually going to agree with you. I think this is a good feature because I, too, get annoyed when I am reading a Substack post and I feel like I am just reading slop, and I kind of want to leave a comment saying, “Did you really write this?” But now I don't have to, because I'll just be able to scan it. But while we are enthusiastic about this feature as Substack readers, some Substack writers seem to be quite upset about it. 404 Media had an article this week in which they quoted at least 1 Substack writer calling this a witch hunt.

Wow.

Now, do you know how you can tell if a witch was created with AI?

How?

6 fingers. Now, you might wonder, why are these writers so concerned about this feature? Well, they're worried about false positives and the reputational damage that they might endure if they are falsely accused of Claude fishing.

Yeah, I am somewhat sympathetic to this critique because we've talked on this show before about how none of these AI detector tools really work perfectly.

Yeah.

I think Pangram is, from what I understand, sort of the best of the lot, but still there are false positives. Some independent studies have found there's some number of false positives where it's accusing people of using AI. And so I am worried about the accuracy of these tools, and I think that there will be maybe a small number, but some number of writers who are falsely accused of using AI because of this feature.

Yeah. You know, this seems like the sort of thing that I should be empathetic about, but I just find that I'm not. And I think ultimately, for most Substackers, it's just not actually going to be that big of a deal to be falsely accused of having used AI, right? These writers, by their nature, are independent writers. These are bloggers. These are not people who are at risk of losing their jobs because they used AI. I think the worst-case scenario here is they have to write a Substack post where they're like, “Hey, I know that this essay kept getting tagged as AI, but it wasn't.”

Yeah.

You know?

I also think that savvy writers will find ways around this. I've actually heard of people running sort of an anti-Pangram loop where they tell an LLM, “Write this essay and then run it through Pangram, and if it keeps coming up as AI-detected, write it again,” and keep doing that until it doesn't come up as AI-detected anymore. So there's sort of a cat-and-mouse game going on here.

I love whenever you do something that just takes way more time than actually writing it yourself. I get very delighted by that. So that seems like a really smart tactic. Other people, though, are saying, Kevin, in addition to the people who are worried about the witch hunt, that there was a writer named Mack Collier who wrote, quote, “I'm not going to apologize for using AI in the creation process. I wrote for 20 years without AI. I could do it again if I wanted to.” But basically he said, “Look, the only thing that AI is doing for me is making my writing better, and why would you not want it to do that?” And I think the answer to that is, “Hey, buddy, go nuts with the AI, but at least for this moment, we want to know, and we want to see the scan.”

Yeah. I have mixed feelings about this because I feel similarly to Mack in the sense that I feel a lot of pressure not to use AI in my writing.

Yeah.

And I do not use AI in my writing—

Yeah.

—to be clear. To write the words on the page or the words in the column, I do not use AI. But I feel there is a lot of pressure, social and otherwise, for writers not to use AI at all during any part of their process.

Yeah.

And I actually feel like journalists and other writers should use the best tools that are available to them. And for some writers, like maybe people who are not—for whom English is not a first language, people who are just starting out—I think this could be excusable, to use AI as a more active part of your writing process. But I do think writers have an obligation for now to disclose this.

Yeah. Well, I think 1 reason why this is all really interesting, Kevin, is trying to understand why Substack felt like they needed to do this, right? And I think the reason is that they must believe that if Substack becomes seen as primarily a destination for slop, it will lower the value of the network, right? Substack makes its money when people go and buy subscriptions, and I think they have rightly intuited people do not want to pay subscriptions to slop. They want to pay subscriptions for human writing. So I'm curious, do you think that is the right bet?

Yeah, I think this is a fair bet for them to make. I think what people are paying for when they pay for a Substack is sort of direct access to a person's brain.

Mm-hmm.

And if what you are getting instead is something that's mostly the output of an LLM, I think people will feel like they're getting a raw deal, like there's some deception going on. And this is an area where I think the norms are going to evolve because pretty soon it will—AI will be integrated in every word processor and content management system, and it will just be kind of unfathomable to do this the old way. But for now, I think there are still enough purists who want 0% AI-generated text in their Substack newsletters that I think this is a good bet for them right now.

Yeah, I think, honestly, there are all sorts of reasons why they might want to do this. I mean, 1 thing about Substack is that anyone can use it for free, right? And even if you have 100,000 free subscribers, they will send an email to 100,000 people for free. That's costing them a lot of money. And I can imagine they're probably staring down the barrel of a future where everyone in every profession is like, “I'm going to start a newsletter. I'm going to have a chatbot write it, a chatbot respond to all of my comments.” And all of a sudden, that's going to become an enormous expense for them. And so they need to find a way to discourage people from doing that sort of thing.

Yeah. I mean, I'll be very curious to see how many people actually use the AI detection tool, because right now, the way it's designed, you have to go out and select this tool. It does not automatically—

Yeah.

—overlay a thing on the post that says, “This was generated by AI.” And my sense is there are a lot of people out there who probably don't care, frankly. If they can't tell that something was generated by AI, what they're reacting to is just, “How good is this thing that I'm reading?” It's not like we've seen all these slop threads go viral on X and other social networks. LinkedIn now is just people posting AI stuff, and my impression is not that it has made LinkedIn a less popular platform. It probably is just people judging it based on the quality of what they're reading.

Well, it's not less popular, but keep in mind, LinkedIn is free to use for the most part, right? LinkedIn isn't making you pay to read those posts, thank God. But Substack is, and so I just think it's different. I will say, when this news was announced, I heard a terrible scream come out of Sand Hill Road down in Silicon Valley as every venture capitalist realized that they could no longer outsource their incessant thought-leadership posts about the economy to Claude anymore, Kevin.

Well, they've been employing human ghostwriters for years to write their tweets and blog posts, so—

And they should do it again.

So we need to keep those jobs afloat.

Now, another interesting dimension of this—and of course this was particularly delicious to me, Kevin, as somebody who dramatically left Substack a few years back because of the Nazis that were on the platform that the company declined to remove—is that in this one moment, Substack has told us that on some level they find AI-generated writing more offensive than Nazi content.

Yes. I think that is one very ungenerous way of interpreting this. I think it is also a threat to their business in a way that maybe Nazi content is not.

Yeah.

It turns out no one actually cares if there's Nazis on your platform.

Yeah.

Well, just you.

Yeah.

It's basically you—

It was me.

You, and three other people. And that's it. But a lot of people do care if their feeds are full of slop.

Here's what people don't realize here in 2026. Nazis used to be very controversial. I would even go so far as to say they were a sort of hated group. But things have changed a lot, and now what we hate is AI.

Yeah.

One more point that I would just make about all of this, and I think this is instructive to everyone on Substack, even if you support this change. This was a really great reminder, I think, and wake-up call that Substack is a platform, and Substack is going to do the things that are best for Substack. And so if you like this particular change, great, but if you don't like it, well, guess what? You're on a platform. The terms can change on you at any time, and they may not be in your favor.

Yeah.

So, yeah.

Have you checked any of the Substacks that you subscribe to using this new AI detection feature?

You know what? I haven't yet. I'm waiting for the next time a VC blog post starts making the rounds to see if I can bust them.

Let's try it now on—

Should we try one now?

Something. Yeah, what should we check?

I went to a blog about AI agents on Substack, which I just assumed was probably generated by AI, and they've disabled the AI detection.

Hmm.

So that's an important thing to point out. As a writer, you can disable AI detection on your blog, but at that point, you might as well just put a big red banner at the top of your website that reads “slop.”

Yeah.

Yeah. In conclusion, I hope that landed well with you. It landed cleanly. That was a load-bearing insights segment. Yeah. Load-bearing insights here on the Hard Fork show. Hard Fork is produced by Whitney Jones and Rachel Cohn. We're edited by Viren Potmange. We're fact-checked by Caitlin Love. Today's show was engineered by Katie McMurran. Original music by Rowan Niemisto, Alyssa Moxley, Chris Wood, and Dan Powell. Video production by Jake Nickell and Chris Schott. You can watch this whole episode on YouTube at youtube.com/hardfork. Special thanks to Paula Schumann, Wee Wing Tam, Brooke Minters, and Dalia Haddad. You can email us at hardfork@nytimes.com with your open letter.