Inside the Mind of Anthropic CEO Dario Amodei | Full documentary | 4K

You know, I've never been someone who slept all that well. Let's just say I'm learning the art of finding ways to relax and sleep through moments of unusual pressure.

It's this feeling of like the exponential. Like suppose you were to accelerate away from Earth on a spaceship at relativistic speed. The way special relativity works is you go to sleep and you wake up and two days have gone by on Earth and so you have to deal with two days in one day. And then you go to sleep and then because you've continued to accelerate three days have gone by on Earth, and then the next day and four days have gone by, and that's a little bit what it feels like.

There are enough clear and present issues that we have to deal with that I'm constantly dealing with those while thinking about how we can prepare. But I don't think paranoia or worrying about what you'll wake up to is productive. I've looked at people in history who've dealt with these very high pressure situations. And you need to learn to respond rationally and not put dangers out of proportion to each other. This yo-yoing between I'm not worried and oh my god we need to panic today — I think that's a hallmark of immature decision-making. And the actual mature decision-making is: you can't ignore this, we can't be complacent, in fact it's getting to be a bigger and bigger risk. But we have to respond rationally, like a surgeon would deal with an operation, or like a military officer would deal with a military operation, or someone making decisions that affect a lot of people has to make those decisions rationally and they have to understand the risk but they have to maintain a basic sense of calm.

We're seeing more and more of them even in the consumer space. We wanted to be more of an enterprise company, but even consumer without us putting that much effort is starting to grow fast.

The interesting thing is that the experience I've had for my whole career and certainly the whole time at Anthropic is that there's this kind of smooth exponential. And the experience of the smooth exponential is: nothing's happening, nothing's happening, nothing's happening, little things happen, and then zoom, it goes crazy. That's the experience of the world. That's the experience of the scale of the company compared to the other companies and compared to the world. So I was watching this graph for a while and I said, 'Oh yeah, we'll probably become the AI company with the most revenue and the most valuation sometime around this time.' And indeed it has happened. So in one sense I'm not surprised because there's just a smooth line on the graph, but of course in another sense when things actually happen you see so much more detail and color to it and it definitely is surprising. But we're just keeping in mind all the things we usually keep in mind which are: how do we train good models? How do we put them in good products? How do we make sure that everything's safe? How do we help people but also manage the societal risks around the technology? It's all the same questions just kind of under a bigger microscope, as it were.

The whole internet revolution was happening around me and I had absolutely no interest in it. I was just interested in doing math and scrolling things. I was interested in understanding the universe. I was interested in science fiction. That was the general milieu. I think I just felt a lot of curiosity about the world.

Yeah, I mean I think the general spirit of nonconformism and individualism, and it's okay to be crazy — I think a good deal of that probably did rub off on me. You hear these stories about you go to countries in Europe or even other parts of this country where it's just kind of discouraged or considered weird to think about things in some different way or have some set of crazy ideas. And there's a lot of things I'm actually very critical about with Silicon Valley, but one thing that I think is good about it is this encouragement of: it doesn't matter if all the experts are against you. It doesn't matter if you have a coherent vision and a coherent worldview — you should go and pursue it. Maybe it just won't work at all. But if it does, there's this kind of long-tailedness to it where there are certain places you can search certain veins where you might find a huge gold mine there. I think that spirit is very important.

That was I think the time when Open Philanthropy Project was first being started up, which Holden was the lead of, and I was at that time a biological scientist. So I was helping them with some of the stuff they were doing around developing world health or biological research. So I kind of advised on that stuff and what were the areas that were promising, what were the areas that were less promising.

Look, I'll say it very simply. There are many difficult issues that you face when you're building powerful technology that Anthropic faces every day where we don't know whether we're making the right decision or the wrong decision. So there are many valid disagreements to be had on safety. We certainly had some of those disagreements with them. But that alone is not sufficient to leave. People here have had disagreements with me. People here have disagreements with each other. But when you feel that you can't trust someone, when you feel that their values are not what they say they are, when you feel that they're not honest, when you feel that they're not in it for the reasons that they say, when you see disturbing patterns of behavior, dishonesty — that makes it very hard to continue to work with a company, to continue to trust the company. And look, at the end of the day, why argue with someone when you don't have the same vision and you don't trust them? The way to resolve it is you go off and do your thing, they go off and do their thing. And I am completely at peace with the idea that we're doing things our way and they're doing things their way. We'll see who wins in the market and we'll see who wins in the court of public opinion. I think those things speak louder than any drama about who left what. We're providing an example of how to deploy this technology in what we think is a responsible way. If they disagree they should make that argument, and I think that's really all there is to say about it.

What happened is that the summit was extremely disorganized. We all came up at the last minute and then they changed the order in which we were standing and then they took a picture of us and then they ordered us all to hold hands. If you've ever been to one of these summits — I'm not saying anything bad about India in particular, but all of these kind of international type summits that have heads of state are super disorganized.

Look, I don't know. I don't know what to tell you. There was like Narendra Modi up there suddenly telling everyone to hold hands.

So here is what I will tell you. There is a wide variance in the quality and the trustworthiness of the people building this technology. I think this meme that no one trusts each other — I don't think it's right. I've known Demis Hassabis, who builds the Gemini models that are a competitor to Claude models, for 15 years. We've worked together on a number of issues. We buy compute from Google. We swap safety ideas all the time. So my view of this is that there are some players who are more trustworthy than others. And I think there are players outside Anthropic who I trust, who I see as trustworthy. What I think needs to happen is that the trustworthy actors need to get together and put the untrustworthy actors in a position where they kind of have to adopt the same standards. I've learned with a lot of experience that there are some folks who don't do the right thing on their own. But if there's a majority of the industry that's doing the right thing, then I think the rest of the industry is kind of left in a position where there's not much they can do but come along. There's the positive version of it where you inspire other people — that's like Demis and me inspiring each other. He does AlphaFold, we're trying to do something in bio as well. We do interpretability research, they start interpretability research. It's not even competition — each company does something cool and the other company's like, 'That's cool, we'd like to do that too and see if there's something new within that we can do.' So that's the carrot side of the race to the top. Then there's the stick side, or the implicit stick, where you're like: okay, these guys are doing the right thing, those guys will look bad if they don't do the right thing. And often we see behaviors where they grudgingly do the right thing while trying to pretend they're doing something different and that there's something bad or sinister about us. That is to be expected. But I think that's the way we get the industry together and that's the way we get the industry to cooperate.

When we started Anthropic, the base thing that mattered, the thing that always matters, is we want to do this right. But then you have to ask yourself, in order to fund the very expensive creation of these models, it needs to be a company that needs to have a business model. Does the business model get in the way of the values? There's always this question, but one of the things I learned — just from being at other companies and watching other companies — is: look, if you pick a business model that fundamentally conflicts with your values, you're going to have a hard time. Either you betray your own values or you become irrelevant. You end up in a catch-22 situation. There are ways out, there are ways to dodge, but it's just a hard situation. It's far better to pick a business model that is compatible with your values. And so when we thought about it, we said: look, we've seen the world of social media, the consumer world — it really seems to encourage engagement, even addiction. The slop we've seen with AI video models — what's going on is they want to maximize the number of minutes that you're paying attention to because that's the advertising revenue-driven incentive. Whereas if we look at enterprise — look, we want to make these models useful to people. If I think of all the positive things you can do with AI — I warn a lot about the negative things, but ultimately we think the positive things will outweigh the negative things — many of those basically fall under the banner of enterprise. We want to use AI to cure diseases that we couldn't cure before. Well, that's working with biotech, working with pharma, working with academic research groups — all of those are enterprises. We want to use AI to make energy cheaper and more efficient — that's all enterprise. We want to use AI to help with education — most of that is enterprise. We want to use AI to address health in the developing world — well, they're nonprofits, but those are basically enterprises. We want to increase economic growth — that is basically enterprise as well. And then there's another factor which is that enterprises care a lot about trust and long-term relationship. Consumer can have this almost gimmicky aspect to it. Where with enterprise, what matters is you build a relationship where you work with a company for many years, you deliver on what you say, they deliver on what they say, and they basically trust you. And so it's very synergistic with our goal of deploying these models in a positive and safe way. And so I think it serves us well to have this business model that largely aligns with our values. Not that there aren't conflicts sometimes, not that there aren't hard choices we have to make, but the number of such choices is much lower than it would be otherwise.

Model quality is the most important thing. We're very far ahead right now on model quality. There is some amount of inertia, but I've never relied on that. Anthropic has never relied on, 'Oh, this is sticky and people won't switch.' I think you want to have a better model. You want to have a better product. And we see the growth rates haven't inflected at all — if anything, they've gone up, at least at the time of taping this interview. So I tend to think that is the most important thing.

Yeah. So this is one of these questions that is kind of very hard to predict in advance. If you could predict it perfectly in advance, then people would and they'd make a huge amount of money on the market and they'd always be right. So no one knows exactly what's going to happen. But I would note a few things. All of these traditional software companies have a number of moats. I think what's going to happen is some of these moats are going to go away but others are going to stay around. The ability to quickly write software — I definitely think that's going away. If your moat is we wrote this complex software that no one else can write, good luck. You're not going to be able to defend that. But I think folks have customer relationships. Folks have know-how of how the field works. Folks have unique domain knowledge. So I think my advice to all of these folks is obviously: don't be complacent, don't ignore it. Make a list of all your moats and be very aware that some of them are going to go away, while others are going to become relatively more important because they're the limiting factors and there may also be new moats. And I think those that definitely respond, that lean into the list of moats that are still present as well as the new ones, will do well. I think those that are complacent, that just delude themselves that what worked in the past will continue to work — they're not going to have a good time. So that is the advice I would give. And I think at the end of the day, I would guess — it depends what you call SaaS and what you don't call SaaS — but I would guess that the software industry gets larger, not smaller, although there will be some big losers.

I just think the pie is getting bigger. With AI, the pie is getting bigger. The existing incumbents may be smaller in relative terms. Some of them may go down in value. Some of them may even go out of business if they don't adapt in the right way. But I think you see this often when growth is really fast. If what's possible with AI grows by 10x, it's very easy for an existing incumbent industry to go up by 1.5x — just not as much as the whole big pie is growing. So I think that may happen. That's not to say we won't have some big losers. I think those who don't adapt, who put their heads in the sand, who don't see what's coming, who don't identify the moats they have — they're going to have a really hard time.

There have been a number of cases where we've really spoken our minds about what we think. I've been very outspoken about the need for export controls on chips to China. I say this because I think it would be really bad for America, for the state of democracy in the world, for China to be ahead in AI capabilities. And some of the chip makers obviously don't agree with that view, but it hasn't stopped me from saying it. I'm saying it again now, even after we've signed more partnerships. What they know is that we always work with them. We've been good partners. We can work together. I'm sure they wish we didn't say these things, but these things are what I believe. What are you going to do? At the end of the day, they benefit from these deals as much as we do. Look, we're all adults here. We can work together on one thing while disagreeing about another thing.

The compute is ramping up very quickly. So it can both be the case that the fundamentals of the business look good but in a year you'll have three times as much compute — three times or four times, or I'm not going to give exact numbers — but these compute ramps are very fast and we have every expectation that the revenue ramp will meet and exceed those. But raising money is kind of the buffer against this cone of uncertainty. So it's a totally rational thing to do. It's a very small dilution to the business. And it logically is not at all the same thing — in fact, it's compatible with the opposite — as knowing that there's anything wrong with the fundamentals of the business.

So one of these things about compute is there's a market in compute. My view is that over a period of time even longer than a couple months, we can get large amounts of compute. One thing that's worth saying here is I don't think we bought too little compute by any reasonable standard. We were planning for a 10x-a-year growth in compute. 10x a year is what we expect. That isn't what we've seen over the first quarter of 2026. We saw a greater than 3x growth in revenue quarterly — just in a quarter, not annualized. Three to the fourth power is 80x over the course of the year. We didn't plan for 80x annualized growth. It would not have been rational to plan for 80x annualized growth, because that means if you only get 10x, you have eight times less. So we're in a locally extreme explosion of compute. That's not going to continue. If that continued, you'd get to revenue numbers that no company on Earth — I don't think that's going to happen, it just can't. But you can have these short periods where it's like, oh my god, this is faster growth than we ever possibly anticipated. But you saw the compute deals with Google, you saw the compute deals with Amazon. There are more that we kind of can and will do. The market's liquid. If you're able to use compute really well and there's the demand, you'll get your compute. It might just take a month or two.

Look, we have a lot of difficult challenges in front of us. There's this race to the top idea that we're trying to pull other companies along with us. And I think we've seen that we have pulled them along with us. Sometimes they don't admit that that's what they're doing. Sometimes they copy us while they're attacking us. But this pull is very valuable. And so I think the value of being the preeminent company both commercially and in terms of models — it's not about beating rivals for the sake of beating rivals. It's about having the ability to pull the ecosystem along with us. And we hope that we can do more of that in the future.

I mean, look, we're always trying to succeed, right? We're not trying to fail here. I'm not someone who believes we should shut this technology down, we shouldn't build it. We exist within a free enterprise system and there's nothing wrong with this. We just have to mitigate the risks of the models. And so it's always been the balance between the two.

What I would say is that I've put a lot of time into thinking about how that's the case. As companies scale, I've been paranoid at every scale. At every scale of the company, there's some new challenge, there's some new way the company can lose either its kind of will to win commercially or the core of its values. I'm worried about both because I see them as synergistic. I actually see the fact that we've been able to make such good models as the thing that allows us to assert our values in a way that works as the company grows, as it gets bigger. There are lots of pitfalls here. There are lots of ways to go wrong. Not because me or the co-founders or the company's leaders' values change, but because the composition of the company changes very fast. So I spend probably half of my time just talking to the company about the culture of Anthropic and how the culture works. When you're growing this fast, you're hiring a bunch of people from big tech companies. If you don't tell them how Anthropic operates, they'll simply recapitulate the only thing they know, which is how to operate at the companies that they came from. And so this is a constant struggle and a constant challenge. And me and Daniela's maybe number one top priority is figuring out how to preserve this, because we recognize that this is the core of who we are in the long run.

I would say two things. The first is, we have a unified company, we have a unified culture. I think we've grown larger while still being incredibly efficient, everyone's still being on the same page — just the cultural and organizational unity. I would say that's the biggest factor. And I would say the second biggest factor is Claude itself — that we're now using Claude to help develop our models and make them more efficient and quickly develop products. There's all kinds of new practices you have to develop. We're still new at it, but it's producing a lot of acceleration and increasingly producing reliable acceleration. And so those are the two factors I would point to.

I think some of the wildest stuff I've seen is around biology and medicine. I've seen a number of cases, including Daniela actually, where Claude diagnosed a medical problem that a bunch of fancy doctors had missed. And on the biology side, the models are starting to get surprisingly good at tasks like drug design or computational chemistry or things like that. And I'm just like, wow — as someone who used to be a biologist I look at it and I'm like, wow, that's hard, you need a lot of training to do that and Claude is getting good at it. That's one area where I think we're going to get a hell of a lot of benefit — that's the positive for AI. We're going to get these huge enormous benefits. Life is going to get better. The quality of human experience is going to get better. A century of scientific progress and a century of progress in what it's like to be human. Go back to 1900. Think of all the problems we had in the 1900s, all the reasons people died prematurely, all the problems they had to suffer, all the material deprivation that we don't have to deal with today. Then think of another hundred years of that. I really believe this century of scientific and medical progress — if we can get through this, and I think we will, I'm increasingly optimistic — we're gonna have a much much better world.

I do. I have not gotten to the point where I actually allow text directly written by Claude in, because I just have such a specific style that I'm a little picky about it. But I basically use Claude to help me brainstorm, to help me think through the themes, to help me kind of figure out what are some references I could use for this. So it kind of plays a supportive role. I don't know how far we are from Claude being able to write better than me. We're not quite there yet, but certainly it's coming.

I'm a little worried about that and in fact that's half the reason I write myself. It certainly is for external audiences — many people read what I write — but it is just as much to clarify my own thinking so that I kind of know what to do next and to create a common reference point across me and others. I think we're still grappling with the question of how exactly do we use AI in a way that kind of preserves those benefits. I think the thing I'm doing now does that, where I use Claude for research and I use Claude for kind of how do I help organize my own thoughts. I think if we just used it end to end — like write an essay about the risks of AI — first of all it wouldn't write the things that I think, but also I would exactly lose that benefit. There's some way as the models get better, I think probably to use them directly, much more directly in the writing, and yet still preserve those benefits. But I think it's going to be a subtle thing. It won't be all one thing. We'll have to kind of figure it out over time.

I think we could have this very unusual combination of very fast GDP growth and high unemployment or at least underemployment or low-wage jobs, a lot of low-wage jobs, high inequality.

I've always said — and if you go back to those original clips they always get cut out of context in like the three seconds — but the real statement was always: I don't know what's going to happen but this is an order of magnitude for how crazy things could be. Also I always talk about all the things we can do in response to this, right? I've talked about token tax and working with enterprises to adjust people and I'm a little skeptical of retraining programs but we should throw them in the mix. Macroeconomic policy. Even from the beginning, I always talked about solutions, but somehow there's this tendency in the human psychology to clip the three seconds of doom is coming. So my message is just definitely not doom is coming. My message is: this is something that we should see coming, that we're worried about, and that we need to actually respond to positively. I don't know exactly, but I'm still pretty concerned. I'm still the same order of concerns. We are seeing right now that AI is making people more productive, but that's the usual hump. If you go back to the kind of industrial revolution — I wrote about this in The Adolescence of Technology — you automate 90% of the job, great, people are 10 times more productive in the other 10% because they're 10 times more leveraged. But eventually it gets close to 100%. Now the sequel to that is: well, then you have to find something else for them to do. I don't know about the long run. I'm truly uncertain about that. But I do think there are types of adaptation. Like one thing I'll talk about is: software engineers within Anthropic — we're going through this transition right now where right now AI makes software engineers more productive even though AI writes all the code or almost all the code. It still makes people more productive. But we're already starting to see the beginning of: there may be some people that it's not making more productive, that it's better for the AI to just do the thing. So that's one side of it. The other side of it is: what do we need more demand for? There's something we call a forward-deployed engineer or an applied AI solutions architect, where their job is a mix of technical work and talking to customers. There's a lot of demand for that because there's a lot of customers and we're growing very quickly. Now, does every person who is in pure software engineering quite work for that? It's not perfect, it's not one-to-one. That gives you a flavor of: there's going to be a hell of a lot of disruption, but things will also adjust. Which wins out, I don't know. But the reason it's important to warn about it is that's how we can respond. That's how we can make policy — both within Anthropic and macroeconomically for the whole world. We want to put out carefully considered thoughts. We don't want to say things that people don't believe will actually do. We don't want to say things that are half-baked. We want to think carefully about what should actually be done about these problems.

So, no one knows for sure, because the economy is unpredictable, right? It's the same as the stock market — they're these kind of decentralized processes that you don't really know ahead of time what are the pieces of the job that people are still going to be able to do. But what I would say broadly is that anywhere that you have these kind of entry-level white collar roles — whether it's banking, whether it's finance — there's going to be a lot of potential for AI to first make people more productive, but then there's going to be a wholesale 'AI can do the job' moment, and then we're going to have to think about: well, what is it that people can do? And I think we need to plan about that ahead of time. We're already doing it. When we talk to enterprise customers, we see choices that they face. They face the choice of: should I save cost, which often means hiring less people — basically do the same thing with less resources — or should we do more things with the same amount of resources? And we always, when we can, try to push them to doing more with the same amount of resources, because basically that means hire the same number of people or maybe even more people, but just do kind of new things — pushing them towards the positive sum. The thing that we have going for us here is the pie is going to expand a lot. And so because the pie is going to expand a lot, there are probably going to be places where people can go. It's just a matter of finding them fast enough. It's the size of the disruption — it's going to be big, and that's what I'm warning people about. But we have to solve that matching problem.

Yeah. No, this is the outcome we want to prevent. This is absolutely the outcome we want to prevent. I think there are a few places — none of them are guaranteed, we're not sure — but there's the physical world, right? Things that are in the physical world. Yes, there's a robotics revolution as well, but it's a lot slower than what's happening in AI. People always talk about building data centers, but when processing information of any type becomes a lot easier, maybe the restriction is going to be things in the physical world. And so we need a lot more people to make, build, manufacture things in the physical world. Anything that's human-centered, I think that's going to be a big deal. People really want to talk to other humans, particularly over important things. Maybe AI can do better customer service, but nevertheless people — or at least some people — want to talk to humans. So these kind of human relationship-driven jobs — I think those are going to be important.

The humans to kind of direct the AIs. Right at some level it has to be in line with someone's values and someone's intentions. And so I think there's going to be some role there, although I don't know how thin versus how thick it will be. I think it's very hard to say.

So I want to be really clear and push back hard against this. The whole picture of there are risks to job loss and here are some ideas. I mean, we haven't fully fleshed out the ideas because I want to get them right. But Anthropic has come up with lots of ideas. We've had economic grants, we have the economic index. I talk about the possible ways to address these risks, from tax and macroeconomic policy to what the new jobs are. In the Adolescence of Technology, I lay out, you know, I have like five pages where I lay out the difference between tasks and jobs, why this time is different than other times, a list of six different things we can do from private philanthropy to government action. I talk about the problems, I talk about the solutions. But social media, which I detest, which I detest as a category, people have these three-second clips from a year ago. They don't actually read the essays, or they prey on the idea that social media. I've written much more carefully about these things where I talk about the risks. The idea that this is cheap marketing is itself cheap marketing. This is laziness. This is failure to engage with serious intellectual work. And I think that is part of the problem. Again, I think it's part of the disease of Silicon Valley. It's been caught up in this social media world of three seconds and so people only respond to it or they think they only have to respond to it. Again, I think it's very dangerous and we fail to have a mature conversation. Instead, people just lazily see this like three-second clip and they're like, 'Oh, this is what Dario was saying.' It's so stupid. It's so unserious. And whenever someone says something like that, I take them less seriously.

Yeah. So what I would say is, look, the world changes. My view of this technology, when I see Russia invading Ukraine, when I see the risk of China invading Taiwan, it worries me that we have a kind of resurgent authoritarian block. They're very aggressive and that we need to defend ourselves. That is something that I have believed for a while now, continue to believe, and that's why across both administrations, you know, I may not agree with every policy of either administration, but that's why we've generally been supportive of this. We don't want a world where China and Russia can analyze all the intelligence with AI, can use AI for attacking Taiwan and Ukraine, and we can't defend them. So that's why we worked with them. We certainly don't do it for the money. It's a huge pain, you know, even putting aside the lawfare, it's just a huge pain to get up on government networks for not that much money. So we did it because we cared about it. But similarly, because we're doing it because we cared about it, there need to be limitations on the use of the technology. And the formulation that I used in Adolescence of Technology: we should use this technology in every way except the ways that undermine our own values, right? And our red lines are mass surveillance and fully autonomous weapons. Those are things that I believe undermine our values. It's not worth democracies winning if democracies do those things. And so that's the balance that I see and that's the stand that we took, and it explains both why we were the first to work with the Department of Defense and why there were some things we wouldn't do when others were willing to do those things. I think you need to pick a stand and stand your ground. This idea of companies that seesaw from 'we won't do anything with the government' to 'suddenly we're doing absolutely everything with the government,' I don't get it. You should pick your principles and stick with them.

We don't work with ICE, either through Palantir or anyone else. We don't work with CBP. I don't believe we work in Gaza. We're very careful about scoping our engagements to things that we believe in.

You know, I don't think there's any winning this fight for a private company. This isn't a fight Anthropic is trying to win or thinks about winning or losing. This is more, I won't even call it a fight. This is more a debate about what the proper use of AI by the government is. And AI is an emerging new technology. We don't understand the ways in which it's reliable or unreliable. We don't understand the ways in which it promotes our values or undermines our values. And so one of the things that I thought was important was to establish a precedent on some of the use cases we think are good, which frankly is most of them, and some of the use cases that we're concerned about. And as I've said, we've already seen, you can only do so much with a contract, right? As we've seen, someone else can sign a contract that doesn't respect your same red lines. But what it has done is raised awareness for the issue. And then we have serious bipartisan efforts in Congress attempting to ban some of the things that we're concerned about and attempting to set guardrails. Again, I don't want to talk about this as a fight, but that's kind of winning—the effort to get our country to think more carefully about what is appropriate use of this technology.

Um, you know, I've been called worse things than that all the time. People can call me or Anthropic whatever they want. The two things that matter are we're successful as a company and we stand up for our values. Like, actually in some ways my life is really easy because when those are the two things you're trying to do, it's really simple. You always know where you stand.

I think there's two things here. There is the ability of the United States to be more effective militarily. I am supportive of that ability. I think having that ability be stronger doesn't cause wars, it deters wars. You know, I basically—you're asking like, do you believe in this country? Do you want this country to be a more powerful actor rather than a less powerful actor on the world stage? I do. I'm a patriot. There's a separate question, which is are there particular policies that the US government is engaged in that I might support or not support. Obviously I support some of them and I don't support others of them. It's not up to me. If we provide a technology, the DoD made this point and we actually agree with them. If we provide a technology, it's not up to us to say you can do this military operation and you can't do that military operation. Now I might privately believe that this military operation makes sense and that military operation is a bad idea, but we're not going to deny the technology. You have to leave policy in the hands of the military decision makers. What you can do is to assert some high-level boundaries that prevent the use cases that seem inconsistent with our values, with our country's values, and promote the use cases that we think encourage our values. So that's how we think about it.

Look, we don't have access to—we don't know exactly how these models were used. Obviously, these things, these mistakes that happen in warfare are really, really terrible. This is a really terrible thing to happen. If that doesn't make clear why we have to stand up for use cases that we don't support, we were willing to risk the future of our company to limit how these models are used. And what you're talking about is a use case that doesn't even violate our red lines. We're worried that there will be a hundred times as much with use cases that do violate our red lines. Now, again, I would say I think overall the use of these models is appropriate, I think it's good on net. But military decision makers make terrible mistakes even at the best of times. And I don't know if we're in the best of times. There are several things we can talk about. We can talk about making red lines that prevent uses of the models that are more likely to lead to those problems, right? If we had allowed, you know, fully autonomous weapons—what we've seen here is Claude assists but a human makes the final call. So a human made that final call, not Claude. Imagine if you had a world in which—not Claude, because we haven't allowed it—but someone else's AI model, the AI model just makes the decision and the human never sees it. That's what we were standing up for. That's what we were fighting against. I would also say, there's a separate thing here. Again, I don't think procurement is the right way to do it, but we need to make sure that it's a matter of interest to the American people. Not to me as a supplier of the technology, but to the American people that our military decision makers don't make these mistakes, that they operate reliably, that they choose wisely what to do. Again, that's of concern to me as a citizen, as a supplier of the technology. The government uses Microsoft Excel a lot. You know, if I said you can use Excel for this military operation but not that, you can't realistically do that. But hopefully that gives you a sense of how we think about it.

Look, what I'm going to say is—and this relies on maybe classified knowledge that I don't have—but the principle that we have established, and I think the principle that was obeyed here, is a human makes the final decision. I don't know what role Claude or any other AI had, but if this isn't an illustration of why that principle is so important, I don't know what is.

I would say on balance it is more likely to stop it. But if we have no limits on how it's used, then I think it could be more likely to cause it. You've seen Dr. Strangelove, right? The premise of it was you have a doomsday device that automatically fires nuclear weapons when it thinks nuclear weapons are being fired at it. What could go wrong? Again, I get to this lethal, fully autonomous weapons thing. I think the way conflicts happen is that the two sides jump at each other, they misunderstand each other. And when we don't have proper oversight of this technology, I think those kinds of accidents are more likely to happen. Now, I think if AI is used in an appropriate way in—not even warfare, but think of just intelligence collection—let's say we're able to predict an invasion of Taiwan or a new movement in Ukraine. Our adversaries will think twice about conducting some kind of invasion or military operation if we know everything that they're doing. And so I think superior intelligence really can deter conflict here. Superior ability to respond can deter conflict. I continue to be a believer in these things.

I think the thing that surprised me most about it was the models had been climbing in their ability to find vulnerabilities and importantly turn those vulnerabilities into exploits, which people only talk about the vulnerabilities. They don't often talk about turning the vulnerabilities into exploits, which it was quite good at. So the things that surprised me are we saw this huge jump. It was a particularly large jump, and without us really prompting them at all, some of the early companies that we gave this to said things like, 'This is a super weapon. You should have to own a gun license to use it. Please don't release this.' The demand to do this was coming from the companies we gave it to, who were finding so many critical vulnerabilities and exploitability around these critical vulnerabilities that they were basically asking us not to release it. Now, to be clear, because things always get distorted in the world of social media, the goal isn't to keep this locked up forever. We're kind of gradually trying to open this up to a wider and wider set of people, and eventually we believe that we should release Mythos to a general audience but with kind of strong cyber safeguards. Now, a concern is today's cyber safeguards, which we did release on Opus 4.7, which is a good cyber model but a substantially weaker one—these can be jailbroken. And we're a little concerned about some of the other companies who think this is a sufficient defense, because it works sometimes, but we all know that these classifiers can be jailbroken or gone around. And our own testing as well as frankly our assessment of the defenses that other companies have put in place suggest that these defenses are not strong enough yet. And that's what we're waiting for—getting the defenses to the point where we really have confidence in them.

The claim that it could be replicated with open-source models, that's just incredibly false. So the idea is Mythos looks across the whole codebase and finds something. Some guy went on Twitter and said, 'Well, if you point an open-source model at exactly the line of code that Mythos finds, then it finds the same issue.' That isn't the point. That isn't the question, right? That is not the same thing. The ultimate test of this is we go to companies, we go to open-source repos. We found 271 new vulnerabilities in Firefox. We've found many thousands within private companies who haven't fixed them yet or can't disclose them yet. Like, no one found those 271 vulnerabilities with the previous models. So the actual workflow of what actually works in practice, as opposed to, 'Okay, I find the exact line that Mythos found, I found the needle in the haystack, something else can now pick up the needle.'

We have suffered enormously commercially from not releasing this model. This model has incredibly accelerated research within Anthropic and production and next models. It would do the same in the outside world if we were to release it. This has hurt us enormously commercially.

What I would say is that the reason we're giving Mythos to defenders before we give it to attackers is to patch all the bugs. As the models get better there may be more and more bugs to be found, but there's only so many—they're finite. It's like you have this surface and there's only so many holes in it. You patch all the holes and then the surface becomes very hard to attack, as well as the code itself is written with the powerful models so it then becomes very hard to find flaws in or break into. So I think on the other side of this, hopefully six months or a year from now, we have a much more secure internet ecosystem than we had in the past. We're trying to get to that world and we're doing the best we can to open up Mythos to new cyber defenders. We've been talking to the government. We're very respectful of their recommendations. They're slowing the pace at which we open it up because they're worried about counterintelligence risk. I think that's sensible. I think all serious people here understand that there's real trade-offs here. We see a lot of sniping from people on Twitter and from other AI companies. You look at what they're saying and the inconsistency with what they're doing. They're not serious people. They're not seriously engaging with the serious trade-offs that we have here. Look, I have customers calling me up every day saying, 'I want access to Mythos.' I have countries calling me up saying, 'I want access to Mythos.' And I have the US government and my security team saying, 'No, wait a minute. There's risk to it.' I'm not saying one side or the other is right. I think it's somewhere in between. Both sides have valid points. But there's a real challenge here and we need to face it together as a society. Not accuse things of being cheap marketing, not use cheap marketing to try and counter-position, which some of the other companies are doing. It just shows an incredible lack of gravitas and maturity. We need to all face this moment together.

Throughout the entire history of Anthropic there have been trade-offs. The entire history of Anthropic, where in some ideal world, you would prefer to, before you release the first chatbot, you could spend years studying every possible thing that could go wrong with it. Now, we did delay the initial release of Claude, but we did it for a few months. So what I'm saying is everything is a trade-off. The extreme ends of the spectrum are completely insane, right? And so everything is a trade-off. What I would say is that now that we're in what I would describe as a commercially leading position, I'm actually, and Daniela are actually doing all we can to move the dial even further towards being careful. That's what the Mythos release was about, right? It's very hard to do something like that if you're not the leading player. And so I think you're going to see more things like that.

So I actually think that's a very serious question and I share those concerns. I don't think the government should outright take us over. But I would put it this way. I would say, just to back up and describe the situation, every previous powerful technology we've seen in history was either built by the government or originated with the government. So nuclear weapons obviously, initially built by the government and pretty much built by the government after that. But even the internet, GPS, cell phones, all the R&D was done in the federal labs and universities. AI is the first technology that's been built in the private sector and where government has not really had a serious role and is coming in late to the game. I think that's actually a dangerous and unstable situation. It is not the situation I would have chosen. There's not really an alternative. This technology is possible to build. Our adversaries are building it. It has economic value. It's going to get built. The issue is the government not doing it, not the private sector doing it. I think we need to think about checks and balances on power. So I think there need to be checks and balances on the power of AI companies. We have this thing, the Long-Term Benefit Trust. What that is, is it's a body that can appoint the majority of the board members and remove the majority of the board members. So it essentially, if you thread it through, has the power to fire me. And what we're looking at is we're introducing some elements—nowhere near all the elements—but we're introducing a little bit of the elements of public governance, where you're accountable to someone who doesn't just have stock in the company. So that's very important and that structure is going to continue no matter what happens to the company. That's on the AI side, and we encourage other companies to have similar structures. On the government side, I think we need checks and balances. There are efforts in Congress that have been announced to enact those red lines. So I really think the legislative branch and the judicial branch need to exert themselves because this technology—I'm scared of companies having it, but I'm also scared of government having it. And then the companies need to provide checks on government and the government needs to provide checks on companies. We need basic regulation of the technology. I think we need to start doing pre-release testing, required pre-release testing, testing and auditing of the models. It's very funny to me how there's a particular group of people in the tech world in Silicon Valley who started with a position of, even having transparency around this technology, even export control, this is all going to apocalyptically destroy our potential to create the technology, it'll kill innovation. And then as soon as they see the first real danger, which I've been expecting all along, there's all this talk of nationalization and the government should just seize it. Come on, folks. You're yo-yoing from the most extreme anti-regulatory, 'if you look at us the wrong way you're destroying the industry,' to this completely communist, 'the government should grab it all.' We need a more sensible, moderate approach. That's the one we've been favoring all along because we've understood the power of this technology. We're not panicking. We're not denying it. We see the smooth exponential and we're responding to it appropriately.

We always try to work together with whoever we can in government. I said we have this simple approach: we have a set of principles, we follow those principles, and we hope that folks on the other side are reasonable. And honestly, the government has taken Mythos very seriously. We've had good conversations with Secretary Bessent, with Chief of Staff Susie Wiles. I think they really understand the nature of the risks here. Mythos has, I think, helped them to feel much more concretely where these risks are. So, again, as with any administration, there are parts who we get along with very well and who understand it, and there are other parts that are harder to get along with. I think that's normal. That would be the case in any administration and we just try to navigate it as best we can.

So one of the things we've seen with this technology is that there's really a premium to how intelligent the models are. We very rarely see that people would prefer to use models with lower intelligence. Now, to be clear, there's a thriving ecosystem. There are lots of challenges and problems that are much easier than the ones we need frontier models for. But again, it's an exponential. It's possible that these far-from-frontier models have economic value comparable to what we saw in 2023 and 2024. But again, we have this 10x a year growth, and so what we find is that what's on the frontier is always much, much larger than what is away from the frontier. I think this is something that people who are used to building products in the previous era don't quite understand. As someone who hadn't run a company before, who's never thought about the previous product era, particularly the social media era, I feel like an outsider to that world and I feel that people's instincts are wrong. They have all these kind of product heuristics and I think the 10x per year model exponential really breaks that. Intelligence is just such a huge factor that it outweighs everything else. And so we're just seeing over and over again that the value is found on the frontier. Now what I do worry about with some of these laggard models is the risks of them, where we have Mythos-class cyber capabilities. Twelve months from now we'll have much better cyber capabilities, but the Mythos-class cyber capabilities may just be available for anyone to download. Now hopefully we'll have patched everything before then. I don't think there's anything we can do to stop it, but I think it's a serious concern.

Not really, no. I worked there for a year. I think I probably learned more about speech recognition and all of that. Maybe the only thing that concerned me was, part of how we got all the speech recognition data was, they said ominously, 'We don't care about privacy in China, so we have all this speech recognition data.' But I think aside from that, my worries here are geopolitical. The things that most worried me about what happened in China are what we saw happen to the Uyghurs, what we saw with suppression of criticism even in the US, with what happened with Hong Kong, right? The fact that the CCP could reach into the US business network and suppress criticism—that's an authoritarian state, a high-tech authoritarian state. And when I see how that combines with AI, you really get a dystopia here, like 1984 or worse. And my focus is on trying to prevent that, and I think we have an opportunity to prevent that. I think we have an opportunity for AI to be a pro-democracy technology that makes people freer, that delivers on the promise of equal justice for all. Or it could go the other way. And which way it goes depends on the actions of the AI companies, it depends on the actions of the government, it depends on the actions of all of us. And so I see us as having responsibility here.

I don't think it's a moment in time. I think it's a continuous process. We're already seeing it in some ways where the AI is able to suggest architectures for the next AI. I would say a year ago we were seeing 10 to 15% kind of increase in total factor productivity due to AI. That's probably up to 20 or 30% now. Might be doubling. As with all things, we're on the exponential. There's no moment where AI improves itself or runs out of control or becomes unsafe. What we have is an accelerating exponential. And at each point on the exponential, we have to assess: is this a time to slow down? Is this a time to put more controls on this technology? I think more and more of that is going to be required. But I think the Rosetta Stone to all of this is the smooth exponential. Again, I think there's an object lesson in the people who were against all AI regulation and then they saw one thing and they wanted to nationalize. I think there's an object lesson in the people who dismissed the power of AI and then said, 'Oh my god, it's improving itself, it's running out of control, we have to shut it all down.' Yo-yoing between those extreme reactions is incredibly unhelpful as a response to this technology. The right response, the wise response, is to say, 'We're not going to panic. Our countermeasures will smoothly ratchet up with the power of the technology.' If you see someone having this kind of crazy yo-yo reaction, that's a sign that they were caught by surprise and that they're not serious.

You know, the figure I most identified with was Leo Szilard, who was the one who first basically had the idea that there could be a kind of chain reaction. Look, my view is we're not going to get through this with larger-than-life personalities or figures who try to be at the center of everything. There needs to be a balance of power here. There's a lot of powerful actors who have interests here and the only way it's going to end well for everyone is if there are basically checks and balances everywhere. So in some ways I actually see Oppenheimer as a failure case, as what should not happen.

I mean, I certainly hope not. My view is that the actions that we have taken lower that probability rather than increasing it. That probability comes from the very straightforward recipe of the technology, the existence of many countries in the world, the existence of many companies within an economy, and new ones created if the void isn't filled. That's a dilemma that we're in. We are trying to act to lower that probability. I think we lower it a lot more than we raise it. But the inherent property of this technology is that it's unpredictable. And so we try to build something and test it a lot before it's released. And then the models that are released today are not dangerous, or at least not really, I think, dangerous outside of cyber. And then we try and iterate and learn from that. There's a zillion defense mechanisms. Half of what we do within the company is try and reduce the risk as much as we can, but it's never going to be zero. I guess what I would say is, suppose there are a bunch of airline companies out there and you're like, 'Well, I'm going to make an airline company that's safer.' It can both be the case that your airline company is 10 times safer than all the other airline companies, but if someone comes and asks you, 'Can you guarantee that your airplane will never crash?' I mean, how could you possibly? But if there was a 25% chance of an airplane crashing, you wouldn't get on that plane.

We're trying to make that probability much, much lower. That is the goal.

My view of this is actually, when any company starts out, and particularly what we've seen with the behavior of just Silicon Valley as an entity, its thinking over the last couple years, I think starting from a position of distrust, you know, if you don't know anything about me, if you don't know anything about Anthropic, is pretty rational. I think Silicon Valley has lost a lot of the world's trust and kind of has to re-earn it. And the message we're trying to send is we're actually different, and that has to be earned in things that we actually do. You can agree or disagree, but we stood up for our values. The thing with Mythos, like it's really hampered us commercially not to put this very powerful model out. And there were a bunch of smaller things before it. We put our money where our mouth is on China. We cut off access to models. We didn't have to do that. No one told us to do that. That cost us several hundred million back when several hundred million was a significant fraction of our revenue. The delay of Claude 2—we have a long history of it. We aren't perfect. We make mistakes. But what I would ask is for people to look at the overall history and say, if you add up that overall history, what is the hypothesis about us that is most consistent with that overall history? People have to decide for themselves. But I think the hypothesis that's consistent is we are genuinely trying to do the right thing. We're imperfect. Organizations are always dysfunctional. We're always trying to fix them and make them work better. Many footfalls, many things that go wrong. But at basis, we have an honest and earnest picture of how to do the right thing. And we're trying to execute on that picture.

Hopefully.

That's one surprising thing that I didn't understand about the CEO job is how often you have to wear makeup. That was not on my bingo card.