Daniela Amodei, Co-Founder and President of Anthropic: Building AI the Right Way | Edited Transcript
A copyedited transcript of Stanford Graduate School of Business's conversation: Daniela Amodei, Co-Founder and President of Anthropic: Building AI the Right Way.
This is a professionally copyedited transcript of Daniela Amodei, Co-Founder and President of Anthropic: Building AI the Right Way. It has been edited for readability and lightly formatted while preserving the substance of the discussion.
Episode Guide
00:00:00 Introduction
00:00:37 From Literature Degree to Silicon Valley
00:05:00 Learning the Language of AI
00:06:52 Starting Anthropic
00:08:56 Choosing Co-Founders You Can Fight With
00:11:31 What AI Safety Actually Means
00:16:40 The Future of Jobs in an AI World
00:19:15 Preparing Society for the AI Transition
00:21:23 AI Adoption Beyond the Silicon Valley
00:24:00 Future With AI
00:27:50 Skills That Will Matter Most
00:30:00 Favorite AI Use Cases
00:33:00 Advice for the Next Generation of Builders
00:35:35 Q&A: Are We in an AI Bubble?
00:38:27 Q&A: Government Regulation and Innovation
00:40:35 Q&A: Privacy and Personal Data
00:43:40 Rapid Fire Questions
00:46:47 Best Advice She’s Ever Received
Made with: The Transcript Desk Chrome Extension
Full video:
In this View From The Top interview, Gintare Zukauskaite, MBA ’26, speaks with Daniela Amodei, Co-Founder and President of Anthropic.
Transcript
00:07
Interviewer: Daniela, welcome to Stanford and View from the Top.
Daniela Amodei: Thank you so much for having me. I’m really excited to be here today.
Interviewer: As you can see, we have a full house. I want to start with the audience: please raise your hand if you use cloud services.
[Many hands and cheers]
Interviewer: Wow, okay! Now, if Sam Altman were here and you asked about ChatGPT, would all of you raise your hands? Let’s be honest.
[Audience laughs]
Interviewer: Still, you’ve got a very supportive crowd here. You and your brother helped build one of the most important AI companies in the world, but neither of you grew up planning for it. Your background is in the arts—you studied English literature—and your early career was in politics. Can you share more about your original career plan?
Daniela Amodei: Oh my God, first of all, thank you for using the word “plan” because I’m not sure I ever really had one. But really, this is a story I hear from a lot of founders or people who end up doing something unexpected. For me, it was about following what interested me most at the time—where my talents, passions, and the chance to make a big impact all met.
Coming out of college in 2009—by the way, not the easiest year to graduate—I had a literature degree and felt like I had no practical skills. I was drawn to making the world better, which has always been a core motivation for both me and my brother, Dari, since we were young. That started with international development and global health, trying to figure out how to build a world where everyone has access to basics like food, water, and medicine.
Even though that’s not directly what I work on now, it gave me a foundation for thinking about how to create something meaningful, something with real purpose that you’re spending a majority of your time on.
From there, it was a winding journey. I worked on Capitol Hill, then on a political campaign, and eventually returned to Silicon Valley. I’m originally from San Francisco, and I started working at Stripe when it was still a tiny company—about 40 people—and no one really knew it. My friends on Capitol Hill were like, “You’re leaving to do what? Payments?” [Laughs] Now it looks like a great decision, but at the time it was a gamble. From there, things snowballed: I went on to work at OpenAI, then co-founded Anthropic.
03:00
Interviewer: You’ve moved across fields seemingly without being limited by your degree or prior experience. Where does that mindset come from—that your background doesn’t have to define your next move?
Daniela Amodei: That’s a great question. I think of myself as a generalist. If you look at my background, you might wonder, “What is she really good at?” I don’t have a law degree or a formal computer science background. But I believe a lot of value lies in being curious and willing to learn across disciplines, paired with a genuine drive to have a positive impact.
That kind of openness and curiosity is an underrated quality. I see it frequently in the people we hire at Anthropic and in the tech industry more broadly. It’s the hallmark of people who are smart, eager to learn, and want to be helpful. Honestly, for most roles aside from engineering at a startup, that’s often more important than the specific degree someone has.
For me, it’s always been about passion and impact—care deeply about what you’re working on, whether intellectually or from a deeper sense of purpose. I was deeply troubled by the unfairness of the world—that just by chance of birth, some people have access to basic necessities, and others don’t.
I wanted to make things fairer and better on a global scale. But I realized I wasn’t having the kind of impact I wanted, so I needed more skills. That’s why I worked on a political campaign; it was inspiring to see how a small, dedicated, driven team can change the world.
Eventually, that path led me to Silicon Valley where you realize you can also make a big difference, but with more resources and maybe a bit less grueling hours compared to politics.
Those core experiences—following your passion, wanting to do more when you care deeply about the work—have been the consistent thread through everything I’ve done.
05:02
Interviewer: Your AI career started when you joined OpenAI in 2018, when it was still a relatively small research lab. Suddenly you were in rooms talking about neural networks, transformers, scaling laws. How did you learn to speak that language?
Daniela Amodei: I had a couple of advantages. First, I spent nearly six years at Stripe working closely with engineers—research and engineering aren’t the same, but there’s definitely some overlap and it gave me a solid grounding.
Second, I grew up with a very talented sibling—my brother, who is a technology-focused physicist—and he’s actually my co-founder at Anthropic. Alongside several others who are very technically skilled, I had a close resource at home to learn from.
05:41-06:50
Daniela Amodei: Anthropic was started by seven of us, all engineers or researchers. What I’d say is that my experiences really instilled in me a sense of not being afraid of technology. Ultimately, it’s just a set of skills—highly valued skills—that anyone can learn. The basics can seem overwhelming at first—there’s a lot of jargon and terminology—but if you keep asking questions and have patient people around you, you can start to understand it. I was lucky to have that support in my life, so I just kept asking questions until I felt comfortable.
The other part was knowing your lane and their lane. There were many things, like training models such as GPT-2 or GPT-3, that the researchers handled, and I probably couldn’t have done that. But I brought other skills to the table that they didn’t have. So understanding your comparative advantage—how you fit into the bigger picture—is really important. That requires interpersonal skills and curiosity, which is innate but also something you can develop. Putting all of that together helped me succeed in that kind of role.
06:53-08:56
Interviewer / Gintare Zukauskaite: In December 2020, you, your brother, and a group of colleagues left OpenAI. Why did you and Dario decide to start Anthropic?
Daniela Amodei: There were seven of us who originally left: me, Dario, and five others, plus some people who joined shortly after. For us, it really came down to focusing on the impact we wanted the technology to have. We’re all people with integrity who care deeply about the impact of what we build. Eventually, it felt easier to create the vision we had outside of the company we were in—by starting something new.
I get asked this question a lot, and the truth is we were running toward something, not away from anything. We had a clear vision: an organization where values like safety and responsibility are front and center. That’s why we incorporated as a public benefit corporation. It took us a while to figure out the best structure because, yes, we’re a commercial entity, and AI will create significant economic value. But it was crucial for us to do it the right way. That shared commitment united the seven of us.
We all had experience working on both AI capabilities and safety and policy work at OpenAI, and it just seemed easier to build this structure anew rather than try to reshape something existing.
08:58-11:28
Interviewer: You’re building Anthropic not only with your brother but with five other co-founders. Many of us here will choose co-founders for the first time, and we all know how often those partnerships end badly. What does it take to make it work?
Daniela Amodei: I consider myself extremely lucky because the seven of us are special on several levels. The first—and this might not be surprising—interpersonal relationships matter way more than you think. How do you handle conflict together? For example, Dario and I have been fighting and making up for almost 40 years—he’s my brother, after all. We’ve known how to work through conflict since childhood. There’s no question we love each other at the end of any fight.
With our co-founders, I’ve known Jared and Chris for about 15 years each; Tom and Sam were roommates; Jared and Sam worked together at Stanford during their PhD. So we had a long history. Dario and I both managed all of the other co-founders at OpenAI, so there was already an established framework for working together, giving feedback, and understanding who we are as people.
Another crucial factor is having a strong, shared vision. If you locked yourself and your co-founder in a room and asked each of you to sketch or write down what you want to build, you shouldn’t end up with one drawing a unicorn and the other a platypus. That kind of mismatch is a sign something’s off.
In our case, we were lucky to come from an environment where we already had this kind of shared focus—we wanted something a little different, but within the same realm. To really test if your partnership will work, do something stressful together—like take a vacation and share a room. If you want to spend more time together afterward, that’s a good sign. If you need a vacation to recover from the vacation, you might want to reconsider.
11:30-11:35
Interviewer: I want to circle back to what you mentioned a moment ago—Anthropic is deeply associated with AI safety, but I want to make sure everyone...
11:36-14:31
Daniela Amodei: I want to make sure everyone understands what AI safety actually means. When people say “AI safety,” it’s a term that’s gotten a bit overused in recent years and often serves as a catch-all for a lot of things. But to us, at a high level, it means taking radical responsibility for the technology we’re building. A good analogy is social media companies—there’s been a lot of criticism of them, and rightly so. But when you look back, the developers behind those platforms didn’t set out to cause problems like teenage girls developing eating disorders. That wasn’t their goal. Instead, they focused on metrics like rapid growth and scale, aiming to build a successful company. At the time, there wasn’t much thought about the massive scale of adoption or the potential consequences.
If you imagine stepping back and saying, “What if I actually tried to anticipate all the ways this could go wrong?” What if I tried to foresee all the unintended side effects in advance and prevent them? Unfortunately, we’re a little privileged in AI because we get to learn from the mistakes of previous tech. Those other companies made errors, and we have the opportunity to say, “We won’t repeat those.” We want to be careful and consider not just how things could go right, but also how things could go wrong.
For us, safety means addressing big risks—like preventing the use of our technology to create chemical or biological weapons, which is a potential danger. It also means dealing with cyber warfare; that’s why we chose not to release our latest model widely due to concerns about cyber misuse. There’s also a lot of work focused on user wellness, child safety, tackling misinformation, and protecting election integrity. This work isn’t new—we stand on the shoulders of teams from other major tech companies who’ve dealt with similar safety and security challenges. We learn from their experience and try to do it better.
14:31-16:41
Interviewer: Anthropic is an AI safety company that also needs to generate revenue. How do you balance those sometimes competing priorities?
Daniela Amodei: We get asked that a lot, and honestly, the tension isn’t as big as people might expect. Most of our revenue comes from businesses, and those companies don’t want unsafe models either. No one’s saying, “Hey, we want Claude to hallucinate more or produce harmful outputs.” So, for the most part, safety and business interests are aligned—being safe is good for business because companies are risk-averse. They want AI that’s reliable and predictable.
However, we are entering a new era where model capabilities are advancing so fast that time becomes the real tension. It’s not that the models can’t do incredible things; it’s that we don’t yet fully understand the risks. So going forward, we’re focused on assessing how serious those risks are and figuring out how to manage them. That sometimes means making unusual decisions, like we did with Project Glasswing. We developed a powerful new class of model that customers would love to use right away, but we felt we needed more time to ensure it was safe before wide release.
That kind of caution is uncomfortable. It’s tough to tell customers, “We want to provide you this tech as soon as possible, but it’s irresponsible to release it before it’s fully patched and safe.” At the end of the day, we come back to our mission. We understand people’s eagerness, but not rushing is the responsible choice.
16:41-17:42
Interviewer: There’s a lot of fear around AI, especially that it will lead to fewer jobs because machines will replace human judgment. Do you think that fear is valid?
Daniela Amodei: It’s a complicated question. AI is definitely going to change the types of jobs available and how people work. We’ve already seen new jobs created in the last five years because of AI, but we’ll also see some jobs disappear over time. According to our economic index—where we study how people currently use AI—the majority of the use cases are complementary. AI is enabling workers rather than replacing them outright. It’s more about augmenting skills and making tasks easier, not just removing human roles.
17:45-18:20
Daniela Amodei: I don’t see AI as a replacement for work, except in a very small number of cases—mostly in customer service. Like, if you have to email Comcast, you probably won’t ever get a human on the other end again. But honestly, I don’t think that’s that different from five years ago.
18:20-19:00
What I really expect is that many types of work will feel somewhat familiar to jobs that exist today, but they won’t be exactly the same. We just don’t know what shape all of that will take. For example, take coding—this is what’s most talked about right now. In business meetings, I hear people say, “We’re discussing Claude,” and then a CEO leans in and asks me conspiratorially, “My daughter’s a sophomore at Stanford—is she still fine to major in computer science, or should she reconsider?” And honestly, we don’t have a clear answer.
My guess is software developers will still be around, but they won’t be writing as much code. A lot of what developers do goes beyond just typing—it’s talking with product managers, working closely with customers. I expect that part of their jobs will expand, while the more routine coding parts might shrink because of AI. But this will open up very different possibilities.
19:00-19:55
What needs to happen next is in education, leadership, and society—we need people to feel prepared and excited about this change, not just anxious. First, we have to approach this with humility—we don’t have all the answers, but we must do the research. At Anthropic, we’ve tried to be radically transparent about these issues, saying upfront that we’re still figuring it out.
Sometimes people say we sound negative when we outline what could happen, but I think it’s better to start the conversation early so no one gets blindsided. For instance, we publish an economic index showing how people are using AI today so everyone can start understanding where this might head. The first step is agreeing, as much as possible, on reality.
Next, we have to be creative and experimental on many levels—how can AI become a grounding and unifying force beyond just work tasks? We need to rethink the connection between work, meaning, and social life because all these will look very different in the future.
19:55-21:15
Lastly, and this is beyond what any tech company can do alone, AI’s impact will become a huge social and political issue. People worry about job displacement already—that anxiety shows up in polling, and it goes beyond just CEOs asking about their kids. We need a broader conversation involving governments, civil society, and universities about what kind of world we want to build as AI takes on more human-like tasks.
21:15-22:00
At Stanford, we live and breathe AI, but Stanford and Silicon Valley aren’t the whole world. What’s blocking AI adoption outside this bubble?
22:00-23:00
Daniela Amodei: That’s a really important point. For us at Anthropic—and probably at Stanford too—it feels like everyone is obsessed with AI. But outside of that circle, many people aren’t comfortable with it yet, and they don’t have the fluency to use these tools well.
Sure, you’ll see impressive stats about AI use, but there’s a demographic element: mostly college-educated people, more men than women, certain racial and wealth groups. Globally, AI adoption isn’t evenly spread.
Interestingly, people in developing countries are usually way more optimistic about AI, seeing it as a huge opportunity and a potential equalizer, while in the US, Europe, and parts of Asia, people tend to feel more anxious. They like how things are and worry about disruption.
23:00-23:39
What we do with this info? I honestly don’t know yet. But it’s clear there are different challenges around access and adoption. We’re still very early in this game, and that’s something that often gets overlooked.
23:40-24:16
Daniela Amodei: One thing I think people tend to miss in Silicon Valley is that we’re in this bubble where it feels like everyone’s already using AI tools, like all the software engineers are using cloud code or CodeX. But that’s not the reality for the vast majority of developers worldwide. So really, the race is just starting. The gun just went off. There’s still so much opportunity to shape how this technology is used and developed, who gets access to it, and what values get baked in. Looking ahead to a future where AI is far more widely adopted…
24:16-24:45
Daniela Amodei: What do we risk losing if we start delegating too much to AI? At Anthropic, we did this huge qualitative survey — I believe the largest ever done — where we spoke with 81,000 people about how they use AI. Some were cloud users, others used different AI tools. What stood out was how varied people’s feelings about AI are, depending on where they are and what they do. Some said, “AI enabled me to do things I never thought I could.”
24:45-25:15
Daniela Amodei: Personally, I didn’t think I’d be able to build a website, but using tools like Cloud, it’s so easy now. I just click a couple of buttons, and voilà! It builds a website for me. Something that might have taken me a year and ended up being a low-quality site now feels effortless. But others express a different feeling. They feel like they don’t have to engage their brain as much anymore.
25:15-25:50
Daniela Amodei: It’s not the same as aimlessly scrolling on your phone. It’s more like, “I could have thought this through or worked on this idea myself, but it’s just easier to trust what the AI spits out.” And I think that’s really where a lot of the anxiety about AI comes from. Humans inherently want to learn, be curious, and expand their horizons. AI can help with that, sure, but if used improperly, it can also disable that impulse.
25:50-26:21
Daniela Amodei: I’ve caught myself thinking, “I could look this up and figure it out myself, but I’ll just ask an AI.” Then I end up blindly trusting what it says — which isn’t always correct. Sometimes Claude gets it wrong. I know it sounds heretical, but it’s true. I think the anxiety lies in how we can set guardrails so that people have to actually make an effort to engage their brains rather than just turning them off.
26:21-27:00
Daniela Amodei: One small example is some of the work we’ve done with universities. We talk about “learning mode,” which some of you might have heard of—maybe you use it, maybe not. When students just put their homework in ChatGPT and get back an answer, that’s cheating — no question. But a different approach is using AI as a patient tutor. For instance, with Claude in learning mode, if you’re stuck writing an essay or something feels off, Claude can help you work through it. It’s like having a personalized professor or tutor who understands what you want to learn and why the class matters to you. It offers to go back over sections, talk things through, help you get unstuck.
27:00-27:46
Daniela Amodei: That’s the kind of interaction where these tools can make you smarter and broaden what you feel capable of learning. But then there’s the other kind of usage where you just turn your brain off and let it do everything for you. I hope as an industry we choose the first approach, not the second.
27:46-28:11
Daniela Amodei: If I had to prioritize which human skills will matter most in an AI-driven world, here’s what I think. Many specific tasks—like being a financial analyst, developer, or copy editor—will change a lot because AI can do many of those tasks.
28:11-29:03
Daniela Amodei: But the real phenomenon is that humans want to be with other humans. We want to spend time together, learn from one another, create together, and really understand each other. We’re social creatures. So in a world where AI handles much of the day-to-day productive work, these social and creative skills will become even more important and highly valued. Because if AI can write code for me, I’m going to want to work with a developer who can explain why something broke or why a tool was designed a certain way.
29:03-29:54
Daniela Amodei: Expanding beyond tech, take medicine as an example: Today, we hire doctors who are great at diagnosing — saying, “Hey, here’s what might be wrong. Let’s run some tests.” AI is going to get really good at that diagnostic part. But what it can’t do is look you in the eye, examine you physically, and understand how you’re feeling emotionally to help you actually feel better. There’s a solid body of medical research showing that...
29:56-30:26
Daniela Amodei: There’s medical research showing that people who have a good relationship with their doctor tend to have better clinical outcomes. It’s hard to explain exactly why, but it’s probably because the doctor puts in that extra effort to really understand what’s wrong. Maybe they run a few unexpected tests or spend more time listening. That kind of bedside manner will become way more important in a world where you’re not just ticking off a list of seven qualifications to decide if a doctor is good enough to treat you.
30:26-31:06
Interviewer (Gintare Zukauskaite): And when you think about the future, are there any AI use cases that you’re personally most excited about?
Daniela Amodei: Oh man. For me personally, I’m a career manager, so I spend a lot of time with people. There’s this weird phenomenon — maybe one day there will be a word for it — where everyone thinks AI isn’t going to take their job because they’re special. I’m totally guilty of this myself. I used to think, people love people; they’re going to want to report to me, not to Claude, our AI system. But actually, Claude is incredibly powerful as a management coach, helping you become a better leader.
31:06-31:26
Daniela Amodei: For example, when we write performance reviews at Anthropic, I’ve uploaded a ton of feedback over three or four years from people who have reported to me. Since people don’t change that much in a few months, Claude’s been really helpful in spotting patterns over time. Like if someone’s been circling around the same issue for several years, maybe they need extra coaching, or someone from the outside to help. It’s the kind of thing that gets missed when you just stay caught in day-to-day work.
31:26-32:07
Daniela Amodei: And in the other direction, Claude is also great at giving me feedback! I upload all my own upward feedback, and Claude gently points out when I haven’t improved in some area over the last year, suggesting maybe I need some additional coaching. I think Claude’s ability to coach and help people be the best versions of themselves has real potential, both at work and in people’s personal lives — as long as it’s handled carefully.
32:07-32:56
Daniela Amodei: Also, I have two little kids — an almost five-year-old and an almost one-year-old — and the best thing Claude’s ever done for me is help me with potty training. That was definitely not a fun experience, but Claude made it a little easier, with empathy and actionable advice. There were even diagrams involved! [laughs] I think Claude’s ability to support overwhelmed parents is huge because there’s so much conflicting information out there. Every time you Google something about your kid, the answer is “Yes, something’s wrong!” Claude offers a more measured and interactive way to help, which I think is really valuable.
33:07-33:19
Interviewer (Gintare Zukauskaite): Before we turn to student questions, Daniela, when you think about the next generation of AI leaders and builders here with us today, what’s the one thing you hope they take away from your journey?
33:19-34:23
Daniela Amodei: Can I say two? Okay, first — this sounds so cliché, I almost don’t want to say it — but honestly, following something you truly care about, something you’re passionate about, is the most important thing you can do. There are so many great ideas out there, but if you don’t feel that burning drive, that sense that this is something the world really needs and that you’d go through walls to make it happen, it’s going to be really hard to keep going when things get tough or boring. There will be tough times — even at Anthropic, there were moments that weren’t fun — but being able to remember why it matters, whether personally or because of the change you want to see in the world, is crucial.
34:23-35:30
Daniela Amodei: Second, for this generation especially, in the last five to ten years, there’s been an understanding that business and doing good don’t have to be at odds with each other. I think that’s a really important and fresh idea. I’m so impressed with this generation of founders and creators who think this way. There’s this marriage between innovation and social impact that feels really special. Stanford has always been exceptional at fostering that. It can’t just be that only “mean, tough” people build businesses — I don’t believe that. Increasingly, I feel that a genuine desire to do good actually correlates really strongly with doing well in business.
35:30-35:56
Interviewer (Gintare Zukauskaite): Great. Let’s turn to the students.
35:56-36:07
Student (Brandon): Hi Daniela, I’m Brandon, a second-year MBA student. Thanks so much for joining us. There’s a debate about whether we’re in an AI bubble, and people usually mean one of three things when they say that: company valuations, how much companies are spending on infrastructure, or whether the pace of AI progress is sustainable. Which of those three are you most...
35:57-38:23
Interviewer: Which of those three risks worry you the most, and which do you think people misunderstand the most?
Daniela Amodei: That’s a great question. I think we need to differentiate some things here. There are different types of risks—some more fragile, some more robust. I wouldn’t say I’m most worried about just one, but one valid concern about this industry is that it’s a high capital expenditure business. That brings inherent risks.
Training these models is extremely expensive. It takes a massive amount of compute power, which is limited in supply. When you combine high demand with scarce supply, prices tend to go up. I’m no economics professor, but that’s basic.
The compute is really the lifeblood of these companies, and you have to buy it far in advance. So, you’re essentially betting on the future—estimating how much compute you’ll need at a certain time. That’s a huge expenditure to make, and it’s a bit nerve-wracking working under those conditions at any of these companies.
Maybe Google is an exception since they’re a public company with deep pockets, but for companies like Anthropic and OpenAI, you’re making a calculated bet that you can pay back that investment over time. We’re obviously very bullish on this space. The revenue potential for these companies is staggering. It’s unlike anything venture capitalists have ever seen before—revenue numbers on this scale in such a short period are almost unimaginable.
If that scenario ever changed drastically, it would cause problems because both Anthropic and OpenAI have already purchased a lot of compute for the future, which is very costly. So, it’s not crazy to be concerned about that risk. We think we’re in a strong position, and the industry overall seems well positioned, but things can change at any time. At the end of the day, it’s still a bet—an industry-wide bet that this will pay off. But, yes, we could absolutely be wrong.
38:23-40:53
Yash (MBA student): Hi Daniela, I’m Yash. What does Anthropic think is the right balance between government regulation and AI innovation? And what do you wish governments worldwide were doing differently?
Daniela Amodei: Great question. This is an area where the conversation has been pretty difficult—largely because the political climate today makes nuanced discussions really hard. It’s a shame because this is a nuanced issue.
I do believe that sensible regulation must play a role in AI’s future. AI is a very different kind of technology than anything that came before it. Honestly, even the last generation of tech probably could have benefited from more thoughtful regulation, in my opinion.
At the same time, I’m not blind to the fact that companies need space to experiment. If you want to create the next generation of incredible products that people want to actually use and adopt, you need room to maneuver.
My biggest hope is that this conversation doesn’t get too politicized, because right now it feels like it’s split into extreme camps: either “regulation is bad, innovation is good,” or “innovation is bad, regulation is good.” But it’s way more complicated than that. Some regulations make sense, and others don’t. There are absolutely critical regulations that ensure technology develops in a way that’s beneficial for people and prevents harm—both to those who rely on AI regularly and society at large.
Ideally, tech companies and regulators should work hand in hand. We have detailed, real-time knowledge of how technology can be misused because we see it in action every day—our safeguards and security teams monitor how people try to poke holes and identify genuine risks. Regulators, meanwhile, know how to create frameworks and systems that can be enforced and followed.
Maybe it’s optimistic, but I hope that we can find common ground where both sides work together. We want to develop amazing technologies—the next Google or Meta or the next breakthroughs no one’s imagined yet—while putting sensible regulations in place that protect people.
40:53-43:00
Jackie Kimmel (MSX student): Hi Daniela, thanks for being here. AI is increasingly accessing sensitive personal data like health records. What do you think individuals should actually be doing to protect their privacy?
Daniela Amodei: That’s another excellent question. You’d be surprised how frequently people casually ask Claude medical questions—it’s actually one of the most common use cases. People often ask things like, “What’s wrong with my kid? What’s wrong with me? Help me.”
There are two sides to this. First, it’s the company’s responsibility to use and protect your data with care. People should absolutely hold companies accountable for how they handle personal data because it’s so sensitive. For example, at Anthropic, we made the conscious decision not to run ads on Claude…
41:55-43:40
Daniela Amodei: Our decision not to put ads in Claude partly comes from the belief that AI technology is fundamentally different. People have conversations with AI tools that are way more personal than what they’d share on Instagram or any social media platform. Because of that, I think technology companies have a greater responsibility to protect your data.
From a personal perspective, I don’t think I have the perfect answer about medical use, but I can tell you many people do use AI for medical questions. I’d urge you to think about safety in terms of not taking AI models at face value when it comes to medical advice. In my own experience, Claude has been correct more often than my doctors on complex medical cases, but I would never act without consulting a licensed medical professional.
We’re very open about the fact that these models sometimes make things up or get confused. They don’t know you personally and they can’t examine you. So, having healthy skepticism is the right approach. Think of it like having a friend who’s a good general doctor—maybe not a specialist—but who can help guide you through your conversation with your actual doctor. Claude is great for that.
It can help you think through things you might not have considered. But my number one recommendation is: don’t blindly follow medical advice from any AI tool. I know you’re all smart enough not to do that, but always check with a professional.
43:40-44:03
Interviewer: Thank you to the students and Daniela. Before we let you go, we have to do our View from the Top rapid-fire tradition.
Daniela: Love it!
44:03-44:09
Interviewer: Are you ready?
Daniela: I’m so ready.
44:09-44:17
Interviewer: Okay, let’s go. What would you major in if you were back in college?
Daniela: [laughs] Can I just say business and get out of this? But… I’d probably major in literature again. I know that sounds crazy, but I love to read.
44:17-44:33
Interviewer: What’s your favorite thing about working with your brother?
Daniela: Oh, Thanksgiving dinner! No, I’d say it’s that we know each other so well. Sometimes we can say things to each other that nobody else in the company could. We kind of get a pass on stuff that no one else would.
44:33-44:54
Interviewer: And what about the least favorite thing about working with your brother?
Daniela: Thanksgiving dinner again, no. I think the hardest part is balancing our personal relationship with our work relationship. We’ve been siblings for a long time before becoming co-founders, and we’ll be siblings long after. So, we make sure to set aside time to hang out outside of work because maintaining that relationship outside the office is really important.
44:54-45:03
Interviewer: Favorite book you found in the library or in your office?
Daniela: Ooh, man. I don’t think I’ve discovered a new one recently, which means I should probably look more. But I was reminded of a favorite — *The Guns of August*. Maybe that counts. It’s about World War I, and it’s a fantastic read if you’re curious about how individual personalities and events led up to the war. It’s a powerful reminder of how so many small decisions and people’s choices ultimately led to something tragic and terrible. I definitely recommend it.
45:03-45:11
Interviewer: Amazing. And if Anthropic had ended up with a different name, what would it be?
Daniela: Oh man. We went through some truly tragic ideas before settling on Anthropic. For some reason, we were really into birds back then—something like Sparrow Systems? No idea where that came from. A bunch of our early model names were birds, like Bert—well, that one wasn’t a bird—but yeah, it was a whole bird thing. Thankfully, better decisions won out and we named it Anthropic, which now feels impossible to imagine as anything else.
45:11-46:00
Interviewer: And finally, not a rapid-fire question: best advice you ever received?
Daniela: Drop the ant! Just kidding. I’m trying to think… Probably something that’s stuck with me is from when we were considering leaving our previous roles to start Anthropic. It didn’t feel obvious then that leaving was the right choice—it felt kind of crazy. I talked to a friend and mentor outside of work, and she said, “Honestly, you don’t need to be on the phone with me. You already know what the right answer is.” And that really stuck with me. Often, when you’re questioning if you’re making the right decision in your life, you actually already know the answer deep down. That was really good advice.
Interviewer: Daniela, it’s been a pleasure.
Daniela: Thank you so much.
Interviewer: Thanks for having me.
47:57
[music]
Made with: The Transcript Desk Chrome Extension