Today, I\u2019m talking with Mike Krieger, the new chief product officer at Anthropic, one of the hottest AI companies in the industry.<\/p>\n<\/div>\n
Anthropic was started in 2021 by former OpenAI executives and researchers who set out to build a more safety-minded AI company \u2014 a real theme among ex-OpenAI employees lately. Anthropic\u2019s main product right now is Claude, the name of both its industry-leading AI model and a chatbot that competes with ChatGPT.\u00a0<\/p>\n<\/div>\n
Anthropic has billions in funding from some of the biggest names in tech, primarily Amazon. At the same time, Anthropic has an intense safety culture that\u2019s distinct among the big AI firms of today. The company is notable for employing some people who legitimately worry AI might destroy mankind<\/a>, and I wanted to know all about how that tension plays out in product design.<\/p>\n<\/div>\n Listen to Decoder<\/em>, a show hosted by The Verge<\/em>\u2019s Nilay Patel about big ideas \u2014 and other problems.\u00a0Subscribe\u00a0here<\/a>!<\/p>\n<\/div>\n<\/div>\n<\/div>\n On top of that, Mike has a pretty fascinating r\u00e9sum\u00e9: longtime tech fans likely know Mike as the cofounder of Instagram, a company he started with Kevin Systrom before selling it to Facebook \u2014 now, Meta \u2014 for $1 billion back in 2012. That was an eye-popping amount back then<\/a>, and the deal turned Mike into founder royalty basically overnight.<\/p>\n<\/div>\n He left Meta in 2018, and a few years later, he started to dabble in AI \u2014 but not quite the type of AI we now talk about all the time on Decoder<\/em>. Instead, Mike and Kevin launched Artifact, an AI-powered news reader that did some very interesting things with recommendation algorithms and aggregation. Ultimately, it didn\u2019t take off like they hoped. Mike and Kevin shut it down earlier this year<\/a> and sold the underlying tech to Yahoo<\/a>.\u00a0<\/p>\n<\/div>\n I was a big fan of Artifact, so I wanted to know more about the decision to shut it down as well as the decision to sell it to Yahoo. Then I wanted to know why Mike decided to join Anthropic and work in AI, an industry with a lot of investment but very few consumer products to justify it. What\u2019s this all for? What products does Mike see in the future that make all the AI turmoil worth it, and how is he thinking about building them?<\/p>\n<\/div>\n I\u2019ve always enjoyed talking product with Mike, and this conversation was no different, even if I\u2019m still not sure anyone\u2019s really described what the future of this space looks like.<\/p>\n<\/div>\n Okay, Anthropic chief product officer Mike Krieger. Here we go.<\/p>\n<\/div>\n This transcript has been lightly edited for length and clarity.\u00a0<\/em><\/p>\n<\/div>\n Mike Krieger, you are the new chief product officer at Anthropic. Welcome to Decoder<\/em>.<\/strong><\/p>\n<\/div>\n Thank you so much. It\u2019s great to be here. It\u2019s great to see you.<\/p>\n<\/div>\n I\u2019m excited to talk to you about products. The last time I talked to you, I was trying to convince you to come to the Code Conference. I didn\u2019t actually get to interview you at Code, but I was trying to convince you to come. I said, \u201cI just want to talk about products with someone as opposed to regulation,\u201d and you\u2019re like, \u201cYes, here\u2019s my product.\u201d<\/strong><\/p>\n<\/div>\n To warn the audience: we\u2019re definitely going to talk a little bit about AI regulation. It\u2019s going to happen. It seems like it\u2019s part of the puzzle, but you\u2019re building the actual products, and I have a lot of questions about what those products could be, what the products are now, and where they\u2019re going.\u00a0<\/strong><\/p>\n<\/div>\n I want to start at the beginning of your Anthropic story, which is also the end of your Artifact story. So people know, you started at Instagram, and you were at Meta for a while. Then you left Meta and you and [Instagram cofounder] Kevin Systrom started Artifact, which was a <\/strong>really fun news reader and had some really interesting ideas about how to surface the web<\/strong><\/a> and have comments and all that, and then you decided to shut it down. I think of the show as a show for builders, and we don\u2019t often talk about shutting things down. Walk me through that decision, because it\u2019s as important as starting things up sometimes.<\/strong><\/p>\n<\/div>\n It really is, and the feedback we\u2019ve gotten post-shutdown for Artifact was some mixture of sadness but also kudos for calling it. I think that there\u2019s value in having a moment where you say, \u201cWe\u2019ve seen enough here.\u201d It was the product I still love and miss, and in fact, I will run into people and I\u2019ll expect them to say, \u201cI love Instagram or I love Anthropic.\u201d They\u2019re always like, \u201cArtifact\u2026 I really miss Artifact.\u201d So clearly, it had a resonance with a too small but very passionate group of folks. We\u2019d been working on the full run of it for about three years, and the product had been out for a year. We were looking at the metrics, looking at growth, looking at what we had done, and we had a moment where we said, \u201cAre there ideas or product directions that will feel dumb if we don\u2019t try before we call it?\u201d\u00a0<\/p>\n<\/div>\n We had a list of those, and that was kind of mid-last year. We basically took the rest of the year to work through those and said, \u201cYeah, those move the needle a little bit,\u201d but it wasn\u2019t enough to convince us that this was really on track to be something that we were collectively going to spend a lot of time on over the coming years. That was the right moment to say, \u201cAll right, let\u2019s pause. Let\u2019s step back. Is this the right time to shut it down?\u201d The answer was yes.<\/p>\n<\/div>\n Actually, if you haven\u2019t seen it, Yahoo basically bought it<\/a>, took all the code, and redid Yahoo News as Artifact<\/a>, or the other way around. It\u2019s very funny. You\u2019ll have a little bit of a Bizarro World moment the first time you see it. You\u2019re like, \u201cThis is almost exactly like Artifact: a little bit more purple, some different sources.\u201d\u00a0<\/p>\n<\/div>\n It was definitely the right decision, and you know it\u2019s a good decision when you step back and the thing you regret is that it didn\u2019t work out, not that you had to make that decision or that you made that exact decision at the time that you did.<\/p>\n<\/div>\n There are two things about Artifact I want to ask about, and I definitely want to ask about what it\u2019s like to sell something to Yahoo in 2024, which is unusual. The first is that Artifact was very much designed to surface webpages. It was predicated on a very rich web, and if there\u2019s one thing I\u2019m worried about in the age of AI, <\/strong>it\u2019s that the web is getting less rich<\/strong><\/a>.<\/strong><\/p>\n<\/div>\n More and more things are moving to closed platforms. More and more creators want to start something new, but they end up on YouTube or TikTok or\u2026 I don\u2019t know if there are dedicated Threads creators yet, but they\u2019re coming. It seemed like that product was chasing a dream that might be under pressure from AI specifically, but also just the rise of creator platforms more broadly. Was that a real problem, or is that just something I saw from the outside?<\/strong><\/p>\n<\/div>\n I would agree with the assessment but maybe see different root causes. I think what we saw was that some sites were able to balance a mix of subscription, tasteful ads, and good content. I would put The Verge<\/em> at the top of that list. I\u2019m not just saying that since I\u2019m talking to you. Legitimately, every time we linked to a Verge<\/em> story from Artifact, somebody clicked through. It was like, \u201cThis is a good experience. It feels like things are in balance.\u201d At the extremes, though, like local news, a lot of those websites for economic reasons have become sort of like, you arrive and there\u2019s a sign-in with Google and a pop-up to sign up to the newsletter before you\u2019ve even consumed any content. That\u2019s probably a longer-run economic question of supporting local news, probably more so than AI. At least that trend seems like it\u2019s been happening for quite a while.<\/p>\n<\/div>\n The creator piece is also really interesting. If you look at where things that are breaking news or at least emerging stories are happening, it\u2019s often an X post that went viral. What we would often get on Artifact is the summary roundup of the reactions to the thing that happened yesterday, which, if you\u2019re relying on that, you\u2019re a little bit out of the loop already. <\/p>\n<\/div>\n When I look at where things are happening and where the conversation is happening, at least for the cultural core piece of that conversation, it\u2019s often not happening anymore on media properties. It is starting somewhere else and then getting aggregated elsewhere, and I think that just has an implication on a site or a product like Artifact and how well you\u2019re ever going to feel like this is breaking news. Over time, we moved to be more interest-based and less breaking news, which, funny enough, Instagram at its heart was also very interest-based. But can you have a product that is just that? I think that was the struggle.<\/p>\n<\/div>\n You said media properties. Some media properties have apps. Some are expressed only as newsletters. But I think what I\u2019m asking about is the web. This is just me doing therapy about the web. What I\u2019m worried about is the web. The creators aren\u2019t on the web. We\u2019re not making websites, and Artifact was predicated on there being a rich web. Search products in general are sort of predicated on there being a rich and searchable web that will deliver good answers.\u00a0<\/strong><\/p>\n<\/div>\n To some extent, AI products require there to be a new web because that\u2019s where we\u2019re training all our models. Did you see that \u2014 that this promise of the web is under pressure? If all the news is breaking on a closed platform you can\u2019t search or index, like TikTok or X, then actually building products on the web might be getting more constrained and might not be a good idea anymore.<\/strong><\/p>\n<\/div>\n Even citing newsletters is a great example. Sometimes there\u2019s an equivalent Substack site of some of the best stuff that I read, and some of the newsletters exist purely in email. We even set up an email account that just ingested newsletters to try to surface them or at least surface links from them, and the design experience is not there. The thing I\u2019ve noticed on the open web in general and as a longtime fan of the web \u2014 somebody who was very online before being online was a thing that people were as a preteen back in Brazil \u2014 is that, in a lot of ways, the incentives have been set up around, \u201cWell, a recipe won\u2019t rank highly if it\u2019s just a recipe. Let\u2019s tell the story about the life that happened leading up to that recipe.\u201d\u00a0<\/p>\n<\/div>\n Those trends have been happening for a while and are already leading to a place where the end consumer might be a user, but it is being intermediated through a search engine and optimized for that findability<\/a> or optimized for what\u2019s going to get shared a bunch or get the most attention. Newsletters and podcasts are two ways that have probably most successfully broken through that, and I think that\u2019s been an interesting direction. <\/p>\n<\/div>\n But in general, I feel like there\u2019s been a decadelong risk for the open web in terms of the intermediation happening between someone trying to tell a story and someone else receiving that story. All the roadblocks along the way just make that more and more painful. It\u2019s no surprise then that, \u201cHey, I can actually just open my email and get the content,\u201d feels better in some ways, although it\u2019s also not great in a bunch of other ways. That\u2019s how I\u2019ve watched it, and I would say it\u2019s not in a healthy place where it is now.<\/p>\n<\/div>\n The way that we talk about that thesis on Decoder<\/em> most often is that people build media products for the distribution. Podcasts famously have open distribution; it\u2019s just an RSS feed. Well, it\u2019s like an RSS feed but there\u2019s Spotify\u2019s ad server in the middle. I\u2019m sorry to everybody who gets whatever ads that we put in here. But at its core, it\u2019s still an RSS product.\u00a0<\/strong><\/p>\n<\/div>\n Newsletters are still, at their core, an IMAP product, an open-mail protocol product. The web is search distribution, so we\u2019ve optimized it to that one thing. And the reason I\u2019m asking this, and I\u2019m going to come back to this theme a few times, is that it felt like Artifact was trying to build a new kind of distribution, but the product it was trying to distribute was webpages, which were already overtly optimized for something else.<\/strong><\/p>\n<\/div>\n I think that\u2019s a really interesting assessment. It\u2019s funny watching the Yahoo version of it because they\u2019ve done the content deals to get the more slimmed-down pages, and though they have fewer content sources, the experience of tapping on each individual story, I think, is a lot better because those have been formatted for a distribution that is linked to some paid acquisition, which is different from what we were doing, which was like, \u201cHere\u2019s the open web. We\u2019ll give you warts and all and link directly to you.\u201d But I think your assessment feels right.<\/p>\n<\/div>\n Okay, so that\u2019s one. I want to come back to that theme. I really wanted to start with Artifact in that way because it feels like you had an experience in one version of the internet that is maybe under pressure. The other thing I wanted to ask about Artifact is that you and Kevin, your cofounder, both once told me that you had big ideas, like scale ideas, for Artifact. You wouldn\u2019t tell me what it was at the time. It\u2019s over now. What was it?<\/strong><\/p>\n<\/div>\n There were two things that I remained sad that we didn\u2019t get to see through. One was the idea of good recommender systems underlying multiple product verticals. So news stories being one of them, but I had the belief that if the system understands you well through how you\u2019re interacting with news stories, how you\u2019re interacting with content, then is there another vertical that could be interesting? Is it around shopping? Is it around local discovery? Is it around people discovery? All these different places. I\u2019ll separate maybe machine learning and AI, and I realize that\u2019s a shifting definition throughout the years, but let\u2019s call it, for the purposes of our conversation, recommender systems or machine learning systems \u2014 for all their promise, my day-to-day is actually not filled with too many good instances of that product.<\/p>\n<\/div>\n The big company idea was, can we bring Instagram-type product thinking to recommender systems and combine those two things in a way that creates new experiences that aren\u2019t beholden to your existing friend and follow graph? With news being an interesting place to start, you highlight some good problems about the content, but the appealing part was that we were not trying to solve the two-sided marketplace all at once. It turns out, half that marketplace was already search-pilled and had its own problems, but at least there was the other side as well. The other piece, even within news, is really thinking about how you eventually open this up so creators can actually be writing content and understanding distribution natively on the platform. I think Substack is pursuing this from a very different direction. It feels like every platform eventually wants to get to this as well.<\/p>\n<\/div>\n When you watch the closest analogs in China, like Toutiao, they started very much with crawling the web and having these eventual publisher deals, and now it is, I would guess, 80 to 90 percent first-party content. There are economic reasons why that\u2019s nice and some people make their living writing articles about local news stories on Toutiao, including a sister or close family member of one of our engineers. But the other side of it is that content can be so much more optimized for what you\u2019re doing.\u00a0<\/p>\n<\/div>\n Actually, at Code, I met an entrepreneur who was creating a new novel media experience that was similar to if Stories met news, met mobile, what would it be for most news stories? I think for something like that to succeed, it also needs distribution that has that as the native distribution type. So the two ideas where I\u2019m like, \u201cone day somebody [will do this]\u201d are recommendation systems for everything and then primarily a recommendation-based first-party content writing platform.\u00a0<\/p>\n<\/div>\n All right, last Artifact question. You shut it down and then there was a wave of interest, and then publicly, one of you said, \u201cOh, there\u2019s a wave of interest, we might flip it,\u201d and then it was Yahoo. Tell me about that process.<\/strong><\/p>\n<\/div>\n There were a few things that we wanted to align. We\u2019d worked in that space for long enough that whatever we did, we sort of wanted to tie a bow around it and move on to whatever it was next. That was one piece. The other piece was that I wanted to see the ideas live on in some way. There were a lot of conversations around, \u201cWell, what would it become?\u201d And the Yahoo one was really interesting, and I would admit to being pretty unaware of what they were doing beyond that I was still using Yahoo Finance in my fantasy football league. Beyond that, I was not familiar with what they were doing. And they were like, \u201cWe want to take it, and we think in two months, we can relaunch it as Yahoo News.\u201d<\/p>\n<\/div>\n I was thinking, \u201cThat sounds pretty crazy. That\u2019s a very short timeline in a code base you\u2019re not familiar with.\u201d They had access to us and we were helping them out almost full time, but that\u2019s still a lot. But they actually pretty much pulled it off. I think it was 10 weeks instead of eight weeks. But I think there is a newfound energy in there to be like, \u201cAll right, what are the properties we want to build back up again?\u201d I fully admit coming in with a bit of a bias. Like, I don\u2019t know what\u2019s left at Yahoo or what\u2019s going to happen here. And then the tech teams bit into it with an open mouth. They went all in and they got it shipped. I\u2019ll routinely text Justin [Bisignano], who was our Android lead and is at Anthropic now. I\u2019ll find little details in Yahoo News, and I\u2019m like, \u201cOh, they kept that.\u201d<\/p>\n<\/div>\n I spent a lot of time with this 3D spinning animation when you got to a new reading level \u2014 it\u2019s this beautiful reflection specular highlighting thing. They kept it, but now it goes, \u201cYahoo,\u201d when you do it. And I was like, \u201cThat\u2019s pretty on brand.\u201d It was a really fascinating experience, but it gets to live on, and it will probably have a very different future than what we were envisioning. I think some of the core ideas are there around like, \u201cHey, what would it mean to actually try to create a personalized news system that was really decoupled from any kind of existing follow graph or what you were seeing already on something like Facebook?\u201d<\/p>\n<\/div>\n Were they the best bidder? Was the decision that Yahoo will deploy this to the most people at scale? Was it, \u201cThey\u2019re offering us the most money\u201d? How did you choose?<\/strong><\/p>\n<\/div>\n It was an optimization function, and I would say the three variables were: the deal was attractive or attractive enough; our personal commitments post-transition were pretty light, which I liked; and they had reach. Yahoo News I think has a hundred million monthly users still. So it was reach, minimal commitment but enough that we felt like it could be successful, and then they were in the right space at least on the bid size.<\/p>\n<\/div>\n It sounds like the dream. \u201cYou can just have this. I\u2019m going to walk away. It\u2019s a bunch of money.\u201d Makes sense. I was just wondering if that was it or whether it wasn\u2019t as much money but that they had the biggest platform, because Yahoo is deceptively huge.<\/strong><\/p>\n<\/div>\n Yeah, it\u2019s deceptively still huge and under new leadership, with a lot of excitement there. It was not a huge exit or I would not call it a super successful outcome, but the fact that I feel like that chapter closed in a nice way and then we could move on without wondering if we should have done something different when we closed it meant that I slept much better at night in Q1 of this year.<\/p>\n<\/div>\n So that\u2019s that chapter. The next chapter is when you show up as the chief product officer at Anthropic. What was that conversation like? Because in terms of big commitments and hairy problems \u2014 are we going to destroy the web? \u2014 it\u2019s all right there, and maybe it\u2019s a lot more work. How\u2019d you make the decision to go to Anthropic?<\/strong><\/p>\n<\/div>\n The top-level decision was what to do next. And I admit to having a bit of an identity crisis at the beginning of the year. I was like, \u201cI only really know how to start companies.\u201d And actually, more specifically, I probably only know how to start companies with Kevin. We make a very good cofounder pair.\u00a0<\/p>\n<\/div>\n I was looking at it like what are the aspects of that that I like? I like knowing the team from day one. I like having a lot of autonomy. I like having partners that I really trust. I like working on big problems with a lot of open space. At the same time, I said, \u201cI do not want to start another company right now. I just went through the wringer on that for three years. It had an okay outcome, but it wasn\u2019t the outcome we wanted.\u201d I sat there saying, \u201cI want to work on interesting problems at scale at a company that I started, but I don\u2019t want to start a company.\u201d<\/p>\n<\/div>\n I kind of swirled a bit, and I was like, \u201cWhat do I do next?\u201d I definitely knew I did not want to just invest. Not that investing is a \u201cjust\u201d thing, but it\u2019s different. I\u2019m a builder at heart, as you all know. I thought, \u201cThis is going to be really hard. Maybe I need to take some time and then start a company.\u201d And then I got introduced to the Anthropic folks via the head of design, who\u2019s somebody I actually built my very first iPhone app with in college. I\u2019ve known him for a long time. His name is Joel [Lewenstein].<\/p>\n<\/div>\n I started talking to the team and realized the research team here is incredible, but the product efforts were so nascent. I wasn\u2019t going to kid myself that I was coming in as a cofounder. The company has been around for a couple of years. There were already company values and a way things were working. They called themselves ants. Maybe I would have advocated for a different employee nickname, but it\u2019s fine. That ship has sailed. But I felt like there was a lot of product greenfield here and a lot of things to be done and built. <\/p>\n<\/div>\n It was the closest combination I could have imagined to 1) the team I would\u2019ve wanted to have built had I been starting a company; 2) enough to do \u2014 so much to do that I wake up every day both excited and daunted by how much there is to do; and 3) already momentum and scale so I could feel like I was going to hit the ground running on something that had a bit of tailwind. That was the combination.<\/p>\n<\/div>\n So the first one was the big decision: what do I do next? And then the second one was like, \u201cAll right, is Anthropic the right place for it?\u201d It was the sort of thing where every single conversation I had with them, I\u2019d be like, \u201cI think this could be it.\u201d I wasn\u2019t thinking about joining a company that was already running like crazy, but I wanted to be closer to the core AI tech. I wanted to be working on interesting problems. I wanted to be building, but I wanted it to feel as close-ish to a cofounder kind of situation as I could. <\/p>\n<\/div>\n Daniela [Amodei], who is the president here, maybe she was trying to sell me, but she said, \u201cYou feel like the eighth cofounder that we never had, and that was our product cofounder,\u201d which is amazing that they had seven cofounders and none of them were the product cofounder. But whatever it was, it sold me, and I was like, \u201cAll right, I\u2019m going to jump back in.\u201d<\/p>\n<\/div>\n I\u2019m excited for the inevitable Beatles documentaries about how you\u2019re the fifth Beatle, and then we can argue about that forever.<\/strong><\/p>\n<\/div>\n The Pete Best event. I hope not. I\u2019m at least the Ringo that comes in later.<\/p>\n<\/div>\n In 2024, with our audience as young as it is, that might be a deep cut, but I encourage everybody to go <\/strong>search for Pete Best<\/strong><\/a> and how much of an argument that is. <\/strong><\/p>\n<\/div>\n Let me ask you two big-picture questions about working in AI generally. You started at Instagram, you\u2019re deep with creatives, you built a platform of creatives, and you obviously care about design. Within that community, AI is a moral dilemma. People are upset about it. I\u2019m sure they\u2019ll be upset that I even talked to you.\u00a0<\/strong><\/p>\n<\/div>\n We had the <\/strong>CEO of Adobe on to talk about Firefly<\/strong><\/a>, and that led to some of the most upset emails we\u2019ve ever gotten. How did you evaluate that? \u201cI\u2019m going to go work in this technology that is built on training against all this stuff on the internet, and people have really hot emotions about that.\u201d There\u2019s a lot to it. There are copyright lawsuits. How did you think about that?<\/strong><\/p>\n<\/div>\n I have some of these conversations. One of my good friends is a musician down in LA. He comes up to the Bay whenever he\u2019s on tour, and we\u2019ll have one-hour conversations over pupusas about AI in music and how these things connect and where these things go. He always has interesting insights on what parts of the creative process or which pieces of creative output are most affected right now, and then you can play that out and see how that\u2019s going to change. I think that question is a big part of why I ended up at Anthropic, if I was going to be in AI.<\/p>\n<\/div>\n Obviously the written word is really important, and there\u2019s so much that happens in text. I definitely do not mean to make this sound like text is less creative than other things. But I think the fact that we\u2019ve chosen to really focus on text and image understanding and keep it to text out \u2014 and text out that is supposed to be something that is tailored to you rather than reproducing something that\u2019s already out there \u2014 reduces some of that space significantly where you\u2019re not also trying to produce Hollywood-type videos or high-fidelity images or sounds and music.\u00a0<\/p>\n<\/div>\n Some of that is a research focus. Some of that is a product focus. The space of thorny questions is still there but also a bit more limited in those domains, or it\u2019s outside of those domains and more purely on text and code and those kinds of expressions. So that was a strong contributor to me wanting to be here versus other spots.<\/p>\n<\/div>\n There\u2019s so much controversy about where the training data comes from. Where does Anthropic\u2019s training data for Claude come from? Is it scraped from the web like everybody else?<\/strong><\/p>\n<\/div>\n [It comes from] scraping the web. We respect robots.txt<\/a>. We have a few other data sources that we license and work with folks separately for that. Let\u2019s say the majority of it is web crawl done in a web crawl respectful way.<\/p>\n<\/div>\n Were you respecting robots.txt before everyone realized that you had to start respecting robots.txt?<\/strong><\/p>\n<\/div>\n We were respecting robots.txt beforehand. And then, in the cases where it wasn\u2019t getting picked up correctly for whatever reason<\/a>, we\u2019ve since corrected that as well.<\/p>\n<\/div>\n What about YouTube? Instagram? Are you scraping those sites?<\/strong><\/p>\n<\/div>\n No. When I think about the players in this space, there are times when I\u2019m like, \u201cOh, it must be nice to be inside Meta.\u201d I don\u2019t actually know if they train on Instagram content or if they talk about that, but there\u2019s a lot of good stuff in there. And same with YouTube. I mean, a close friend of mine is at YouTube. That\u2019s the repository of collective knowledge of how to fix any dishwasher in the world, and people ask that kind of stuff. So we\u2019ll see over time what those end up looking like.<\/p>\n<\/div>\n You don\u2019t have a spare key to the Meta data center or the Instagram server?<\/strong><\/p>\n<\/div>\n [Laughs] I know, I dropped it on the way out.<\/p>\n<\/div>\n When you think about that general dynamic, there are a lot of creatives out there who perceive AI to be a risk to their jobs or perceive that there\u2019s been a big theft. I\u2019ll just ask about the <\/strong>lawsuit against Anthropic<\/strong><\/a>. It\u2019s a bunch of authors who say that Claude has illegally trained against their books. Do you think there\u2019s a product answer to this? This is going to lead into my second question, but I\u2019ll just ask broadly, do you think you can make a product so good that people overcome these objections?<\/strong><\/p>\n<\/div>\n Because that is kind of the vague argument I hear from the industry. Right now, we\u2019re seeing a bunch of chatbots and you can make the chatbot fire off a bunch of copyrighted information, but there\u2019s going to come a turn when that goes away because the product will be so good and so useful that people we\u2019ll think it has been worth it. I don\u2019t see that yet. I think a lot of the heart of the copyright lawsuits beyond just the legal piece of it is that the tools are not so useful that anyone can see that the trade is worth it. Do you think there\u2019s going to be a product where it is obvious that the trade is worth it?<\/strong><\/p>\n<\/div>\n I think it\u2019s very use case dependent. The kind of question that we drove our Instagram team insane with is we would always ask them, \u201cWell, what problem are you solving?\u201d A general text bot interface that can answer any question is a technology and the beginnings of a product, but it\u2019s not a precise problem that you are solving. Grounding yourself in that maybe helps you get to that answer. For example, I use Claude all the time for code assistance. That is solving a direct problem, which is, I\u2019m trying to ramp up on product management and get our products underway and also work on a bunch of different things. To the extent that I have any time to be in pure build mode, I want to be really efficient. That is a very directly connected problem and a total game-changer just through myself as a builder, and it lets me focus on different pieces as well.<\/p>\n<\/div>\n I was talking to somebody right before this call. They are now using Claude to soften up or otherwise change their long missives on Slack before they send them. This kind of editor solves their immediate problem. Maybe they need to tone it down and chill out a little bit before sending a Slack message. Again, that grounds it in use because that\u2019s what I\u2019m trying to really focus on. If you try to boil the ocean, I think you end up really adjacent to these kinds of ethical questions that you raise. If you\u2019re an \u201canything box,\u201d then everything is potentially either under threat or problematic. I think there\u2019s real value in saying, \u201cAll right, what are the things we want to be known to be good for?\u201d<\/p>\n<\/div>\n I\u2019d argue today that the product actually does serve some of those well enough that I\u2019m happy it exists and I think folks are in general. And then, over time, if you look at things like writing assistance more broadly for novel-length writing, I think the jury\u2019s still out. My wife was doing kind of a prototype version of that. I\u2019ve talked to other folks. Our models are quite good, but they\u2019re not great at keeping track of characters over book-length pieces or reproducing particular things. I would ground that in \u201cwhat can we be good at now?\u201d and then let\u2019s, as we move into new use cases, navigate those carefully in terms of who is actually using it and make sure we\u2019re providing value to the right folks in that exchange.<\/p>\n<\/div>\n Let me ground that question in a more specific example, both in order to ask you a more specific question and also to calm the people who are already drafting me angry emails.<\/strong><\/p>\n<\/div>\n TikTok exists. TikTok is maybe the purest garden of innovative copyright infringement that the world has ever created. I\u2019ve watched entire movies on TikTok, and it\u2019s just because people have found ways to bypass their content filters. I do not perceive the same outrage at TikTok for copyright infringement as I do with AI. Maybe someone is really mad. I have watched entire 1980s episodes of This Old House<\/em> on TikTok accounts that are labeled, \u201cBest of This Old House.\u201d I don\u2019t think Bob Vila is getting royalties for that, but it seems to be fine because TikTok, as a whole, has so much utility, and people perceive even the utility of watching old 1980s episodes of This Old House<\/em>.<\/strong><\/p>\n<\/div>\n There\u2019s something about that dynamic between \u201cthis platform is going to be loaded full of other people\u2019s work\u201d and \u201cwe\u2019re going to get value from it\u201d that seems to be rooted in the fact that, mostly, I\u2019m looking at the actual work. I\u2019m not looking at some 15th derivative of This Old House<\/em> as expressed by an AI chatbot. I\u2019m actually just looking at a 1980s version of This Old House<\/em>. Do you think that AI chatbots can ever get to a place where it feels like that? Where I\u2019m actually looking at the work or I\u2019m providing my attention or time or money to the actual person who made the underlying work, as opposed to, \u201cWe trained it on the open internet and now we\u2019re charging you $20, and 15 steps back, that person gets nothing.\u201d<\/strong><\/p>\n<\/div>\n To ground it in the TikTok example as well, I think there\u2019s also an aspect where if you imagine the future of TikTok, most people probably say, \u201cWell, maybe they\u2019ll add more features and I\u2019ll use it even more.\u201d I don\u2019t know what the average time spent on it is. It definitely eclipses what we ever had on Instagram.\u00a0<\/p>\n<\/div>\n That\u2019s terrifying. That\u2019s the end of the economy.<\/strong><\/p>\n<\/div>\n Exactly. \u201cBuild AGI, create universal prosperity so we can spend time on TikTok\u201d would not be my preferred future outcome, but I guess you could construct that if you wanted to. I think the future feels, I would argue, a bit more knowable in the TikTok use case. In the AI use case, it\u2019s a bit more like, \u201cWell, where does this accelerate to? Where does this eventually complement me, and where does it supersede me?\u201d I would posit that a lot of the AI-related anxiety can be tied to the fact that this technology was radically different three or four years ago.<\/p>\n<\/div>\n Three or four years ago, TikTok existed, and it was already on that trajectory. Even if it weren\u2019t there, you could have imagined it from where YouTube and Instagram were. If they had an interesting baby with Vine, it might\u2019ve created TikTok. It is partially because the platform is so entertaining; I think that\u2019s a piece. That connection to real people is an interesting one, and I\u2019d love to spend more time on that because I think that\u2019s an interesting piece of the AI ecosystem. Then the last piece is just the knowability of where it goes. Those are probably the three [elements] that ground it more.\u00a0<\/p>\n<\/div>\n Anthropic started, it was probably the original \u201cwe\u2019re all quitting OpenAI to build a safer AI\u201d company. <\/strong>Now there are a lot of them<\/strong><\/a>. My friend Casey [Newton] makes a joke that every week someone quits to start yet another safer AI company. Is that expressed in the company? Obviously Instagram had big moderation policies. You thought about it a lot. It is not perfect as a platform or a company, but it\u2019s certainly at the core of the platform. Is that at the core of Anthropic in the same way that there are things you will not do?<\/strong><\/p>\n<\/div>\n Yes, deeply. And I saw it in week two. So I\u2019m a ship-oriented person. Even with Instagram\u2019s early days, it was like, \u201cLet\u2019s not get bogged down in building 50 features. Let\u2019s build two things well and get it out as soon as possible.\u201d Some of those decisions to ship a week earlier and not have every feature were actually existential to the company. I feel that in my bones. So week two, I was here. Our research team put out a paper on interpretability of our models, and buried in the paper was this idea that they found a feature inside one of the models that if amplified would make Claude believe it was the Golden Gate Bridge<\/a>. Not just kind of believe it, like, as if it were prompted, \u201cHey, you\u2019re the Golden Gate Bridge.\u201d [It would believe it] deeply \u2014 in the way that my five-year-old will make everything about turtles, Claude made everything about the Golden Gate Bridge.<\/p>\n<\/div>\n \u201cHow are you today?\u201d \u201cI\u2019m feeling great. I\u2019m feeling International Orange and I\u2019m feeling in the foggy clouds of San Francisco.\u201d Somebody in our Slack was like, \u201cHey, should we build and release Golden Gate Claude?\u201d It was almost an offhand comment. A few of us were like, \u201cAbsolutely yes.\u201d I think it was for two reasons. One, this was actually quite fun, but two, we thought it was valuable to get people to have some firsthand contact with a model that has had some of its parameters tuned. From that IRC message to having Golden Gate Claude out on the website was basically 24 hours. In that time, we had to do some product engineering, some model work, but we also ran through a whole battery of safety evals.\u00a0<\/p>\n<\/div>\n That was an interesting piece where you can move quickly, and not every time can you do a 24-hour safety evaluation. There are lengthier ones for new models. This one was a derivative, so it was easier, but the fact that that wasn\u2019t even a question, like, \u201cWait, should we run safety evals?\u201d Absolutely. That\u2019s what we do before we launch models, and we make sure that it\u2019s both safe from the things that we know about and also model out what some novel harms are. The bridge is unfortunately associated with suicides. Let\u2019s make sure that the model doesn\u2019t guide people in that direction, and if it does, let\u2019s put in the right safeguards. Golden Gate Claude is a trivial example because it was like an Easter egg we shipped for basically two days and then wound down. But [safety] was very much at its core there. <\/p>\n<\/div>\n Even as we prepare model launches, I have urgency: \u201cLet\u2019s get it out. I want to see people use it.\u201d Then you actually do the timeline, and you\u2019re like, \u201cWell, from the point where the model is ready to the point where it\u2019s released, there are things that we are going to want to do to make sure that we\u2019re in line with our responsible scaling policy.\u201d I appreciate that about the product and the research teams here that it\u2019s not seen as, \u201cOh, that\u2019s standing in our way.\u201d It\u2019s like, \u201cYeah, that\u2019s why this company exists.\u201d I don\u2019t know if I should share this, but I\u2019ll share it anyway. At our second all-hands meeting since I was here, somebody who joined very early here stood up and said, \u201cIf we succeeded at our mission but the company failed, I would see this as a good outcome.\u201d\u00a0<\/p>\n<\/div>\n I don\u2019t think you would hear that elsewhere. You definitely would not hear that at Instagram. If we succeeded in helping people see the world in a more beautiful, visual way, but the company failed, I\u2019d be super bummed. I think a lot of people here would be very bummed, too, but that ethos is quite unique.<\/p>\n<\/div>\n This brings me to the Decoder<\/em> questions. <\/strong>Anthropic is what\u2019s called a public benefit corporation<\/strong><\/a>. There\u2019s a trust underlying it. You are the first head of product. You\u2019ve described the product and research teams as being different, then there\u2019s a safety culture. How does that all work? How is Anthropic structured?<\/strong><\/p>\n<\/div>\n I would say, broadly, we have our research teams. We have the team that sits most closely between research and product, which is a team thinking about inference and model delivery and everything that it takes to actually serve these models because that ends up being the most complex part in a lot of cases. And then we have product. If you sliced off the product team, it would look similar to product teams at most tech companies, with a couple of tweaks. One is that we have a labs team, and the purpose of that team is to basically stick them in as early in the research process as possible with designers and engineers to start prototyping at the source, rather than waiting until the research is done. I can go into why I think that\u2019s a good idea. That\u2019s a team that got spun up right after I joined.<\/p>\n<\/div>\n Then the other team we have is our research PM teams, because ultimately we\u2019re delivering the models using these different services and the models have capabilities, like what they can see well in terms of multimodal or what type of text they understand and even what languages they need to be good at. Having end-user feedback tied all the way back to research ends up being very important, and it prevents it from ever becoming this ivory tower, like, \u201cWe built this model, but is it actually useful?\u201d We say we\u2019re good at code. Are we really? Are startups that are using it for code giving us feedback on, \u201cOh, it\u2019s good at these Python use cases, but it\u2019s not good at this autonomous thing\u201d? Great. That\u2019s feedback that\u2019s going to channel back in. So those are the two distinct pieces. Within product, and I guess a click down, because I know you get really interested on Decoder<\/em> about team structures, we have apps, just Claude AI, Claude for Work, and then we have Developers, which is the API, and then we have our kooky labs team.<\/p>\n<\/div>\n That\u2019s the product side. The research side, is that the side that works on the actual models?<\/strong><\/p>\n<\/div>\n Yeah, that\u2019s the side on the actual models, and that\u2019s everything from researching model architectures, to figuring out how these models scale, and then a strong red teaming safety alignment team as well. That\u2019s another component that is deeply in research, and I think some of the best researchers end up gravitating toward that, as they see that\u2019s the most important thing they could work on.<\/p>\n<\/div>\n How big is Anthropic? How many people?<\/strong><\/p>\n<\/div>\n We\u2019re north of 700, at last count.<\/p>\n<\/div>\n And what\u2019s the split between that research function and the product function?<\/strong><\/p>\n<\/div>\n Product is just north of 100, so the rest is everything between: we have sales as well, but research, the fine-tuning part of research, inference, and then the safety and scaling pieces as well. I described this within a month of joining as those crabs that have one super big claw. We\u2019re really big on research, and product is still a very small claw. The other metaphor I\u2019ve been using is that, you\u2019re a teenager, and some of your limbs have grown faster than others and some are still catching up. <\/p>\n<\/div>\n The crazier bet is that I would love for us to not have to then double the product team. I\u2019d love for us instead to find ways of using Claude to make us more effective at everything we do on product so that we don\u2019t have to double. Every team struggles with this so this is not a novel observation. But I look back at Instagram, and when I left, we had 500 engineers. Were we more productive than at 250? Almost certainly not. Were we more productive than at 125 to 250? Marginally? <\/p>\n<\/div>\n I had this really depressing interview once. I was trying to hire a VP of engineering, and I was like, \u201cHow do you think about developer efficiency and team growth?\u201d He said, \u201cWell, if every single person I hire is at least net contributing something that\u2019s succeeding, even if it\u2019s like a 1 to 1 ratio\u2026\u201d I thought that was depressing. It creates all this other swirl around team culture, dilution, etc. That\u2019s something I\u2019m personally passionate about. I was like, \u201cHow do we take what we know about how these models work and actually make it so the team can stay smaller and more tight-knit?\u201d<\/p>\n<\/div>\n Tony Fadell, who did the iPod, <\/strong>he\u2019s been on Decoder<\/em> before<\/strong><\/a>, but when we were starting The Verge<\/em>, he was basically like, \u201cYou\u2019re going to go from 15 or 20 people to 50 or 100 and then nothing will ever be the same.\u201d I\u2019ve thought about that every day since because we\u2019re always right in the middle of that range. And I\u2019m like, when is the tipping point?\u00a0<\/strong><\/p>\n<\/div>\n Where does moderation live in the structure? You mentioned safety on the model side, but you\u2019re out in the market building products. You\u2019ve got what sounds like a very horny Golden Gate Bridge people can talk to \u2014 sorry, every conversation has one joke about how horny the AI models are.<\/strong><\/p>\n<\/div>\n [Laughs<\/em>] That is not what that is.<\/p>\n<\/div>\n Where does moderation live? At Instagram, there\u2019s the big centralized Meta trust and safety function. At YouTube, <\/strong>it\u2019s in the product org under Neal Mohan<\/strong><\/a>. Where does it live for you?<\/strong><\/p>\n<\/div>\n I would broadly put it in three places. One is in the actual model training and fine-tuning, where part of what we do on the reinforcement learning side is saying we\u2019ve defined a constitution for how we think Claude should be in the world<\/a>. That gets baked into the model itself early. Before you hit the system prompt, before people are interacting with it, that\u2019s getting encoded into how it should behave. Where should it be willing to answer and chime in, and where should it not be? That\u2019s very linked to the responsible scaling piece. Then next is in the actual system prompt. In the spirit of transparency, we just started publishing our system prompts<\/a>. People would always figure out clever ways to try to reverse them anyway, and we were like, \u201cThat\u2019s going to happen. Why don\u2019t we just actually treat it like a changelog?\u201d\u00a0<\/p>\n<\/div>\n As of this last week, you can go online and see what we\u2019ve changed. That\u2019s another place where there\u2019s additional guidance that we give to the model around how it should act. Of course, ideally, it gets baked in earlier. People can always find ways to try to get around it, but we\u2019re fairly good at preventing jailbreaks. And then the last piece is where our trust and safety team sits, and the trust and safety team is the closest team. At Instagram, we called it at one point trust and safety, at another point, well-being. But it\u2019s that same kind of last-mile remediation. I would bucket that work into two pieces. One is, what are people doing with Claude and publishing out to the world? So with Artifacts, it was the first product we had that had any amount of social thing at all, which is that you could create an Artifact, hit share, and actually put that on the web<\/a>. That\u2019s a very common problem in shared content.<\/p>\n<\/div>\n I lived shared content for almost 10 years at Instagram, and here, it was like, \u201cWait, do people have usernames? How do they get reported?\u201d We ended up delaying that launch by a week and a half to make sure we had the right trust and safety pieces around moderation, reporting, cues around taking it down, limited distribution, figuring out what it means for the people on teams plans versus individuals, etc. I got very excited, like, \u201cLet\u2019s ship this. Sharing Artifacts.\u201d Then, a week later, \u201cOkay, now we can ship it.\u201d We had to actually sort those things out.<\/p>\n<\/div>\n So that\u2019s on the content moderation side. And then, on the response side, we also have additional pieces that sit there that are either around preventing the model from reproducing copyrighted content, which is something that we want to prevent as well from the completions, or other harms that are against the way we think the model should behave and should ideally have been caught earlier. But if they aren\u2019t, then they get caught at that last mile. Our head of trust and safety calls it the Swiss cheese method, which is like, no one layer will catch everything, but ideally, enough layer stack will catch a lot of it before it reaches the end.<\/p>\n<\/div>\n