Dario Amodei:
Who Is the Father of Claude
FROM ANTHROPIC TO A SAFER AI
Dario Amodei is a scientist who quickly grasped the importance of technological progress for humanity. A progress that, however, must be guided by the values that set this entrepreneur apart from his Silicon Valley peers. It’s with this awareness that he founded his company, Anthropic, and conceived an LLM of his own, Claude, becoming in the process a figure many find uncomfortable. Let’s find out the story of a special, unconventional man who chose an ethical path to face a complex future.
( by: Antonio Maria Guerra | date: 13/05/2026 )
Dario Amodei: an unconventional entrepreneur.
Despite being a prominent figure in contemporary Artificial Intelligence, Dario Amodei remains largely unknown to the general public.
A physicist by training, a researcher by passion, and an entrepreneur by deliberate choice, he is the co-founder and CEO of Anthropic (*1), the company behind Claude, the AI system widely recognized and respected in the world.
Born in San Francisco in 1983 to a father of Italian origin and an American mother, his story is one of scientific rigor combined with remarkable intellectual ambition and, above all, a steadfast commitment to his principles, a quality that sets him apart from his Silicon Valley counterparts. His path has never been driven by commercial success, but by the desire to understand how complex systems, first biological and then artificial, can be comprehended, guided, and made safe. This deep-seated interest took him from biophysics research at Princeton to advanced work on large language models (LLMs) at OpenAI, where he played a central role in the development of the now-celebrated GPT-2 and GPT-3 (*2).
Read more
His next step was, in retrospect, entirely foreseeable: he founded his own company, Anthropic, to prove that power and safety are not contradictory goals.
It comes as no surprise, then, that TIME included Amodei among the hundred influential people in the world in both 2025 and 2026 (*3).
Far from being an enthusiastic techno-optimist or a catastrophist doomer, this man stands at the center of the most consequential technology in human history, clear-eyed about everything that could go wrong, yet guided by a precise vision of everything that could go extraordinarily right.
Note:
*1: darioamodei.com;
*2: Wikipedia, “Dario Amodei”;
*3: TIME, TIME100 2025 e TIME100 2026;
PODCAST: Dario Amodei, Who Is the Father of Claude.
The origins of the name ‘Anthropic’.
‘Anthropic’, the name Dario Amodei gave his company, or rather ‘his child’, is no coincidence: it evokes, in fact, the ‘anthropic principle’, a concept in cosmology holding that the universe, as we know it, is inevitably shaped by the very existence of the human beings who inhabit it (*1). This choice, on reflection, reveals much about the man: a scientist first, and a businessman second.
Note
*1 Stanford Encyclopedia of Philosophy, “Anthropic Principle”.
San Francisco, the 1980s and ‘90s: Italian roots, between two worlds.
Born in 1983 in San Francisco’s Mission District, Dario Amodei grows up as the son of two apparently distant worlds: that of his father Riccardo, a leather craftsman of Tuscan origin from Massa Marittima, and that of his mother Elena, an American woman of Jewish heritage who managed library construction and renovation projects (*1). As a child, he observes both: on one side, the patient, hands-on work of the former; on the other, the methodical precision of the latter, who classifies, orders, and plans. In the middle stands a curious boy trying to make sense of a universe populated by adults where everyone has opinions and no one ever seems to be right or wrong.
Read more
He finds his path in mathematics and science: fields where opinions carry no weight and no gray areas exist. There is only one truth, waiting to be discovered. This fascination defines his high school years, while around him the dot-com boom explodes and Silicon Valley seems to promise fortunes to anyone who knows how to build a website. But Dario is not interested: he would later recall that the idea of creating web pages never appealed to him (*2). His attention was already elsewhere, drawn to complex systems, the laws that govern reality, and everything that lies beneath the surface of things. A calling destined to leave a profound mark on his future.
Notes:
*1: Wikipedia, “Dario Amodei”, Founders File;
*2: Alex Kantrowitz, “The Making of Anthropic CEO Dario Amodei”, Medium, July 2025;
Early 2000s: progress as a ‘moral duty’.
There is no point denying it: some events permanently change the way people see the world, and for Dario Amodei one of them arrives when he is barely in his twenties, with the illness that strikes his father Riccardo. While those around him seek comfort in hope, the young man falls back on what he does best: looking for a rational answer. He reads scientific articles, clinical studies, and medical documentation, searching for a possible cure (*1). But fate has other plans: his father does not make it. Only a few years later, a new therapy emerges for the same illness, with unexpected results: a milestone in science that leaves Dario with one bitter, unanswerable question: “What if it had been discovered sooner?”.
Read more
From that moment on, he carries with him a constant sense that time matters more than anything else, and that technological progress is not a mere intellectual exercise, but a factor that can, in very concrete terms, determine life or death.
Building something useful ceases to be a career choice, becoming instead a genuine moral duty.
Note:
*1: Wikipedia, “Dario Amodei”;
Dario Amodei: an early talent.
At Lowell High School, one of San Francisco’s most selective public schools, the young Amodei does not go unnoticed. In 2000, he joins the American team for the International Physics Olympiad (*1), a competition that measures the finest scientific talents on the planet. This is no minor detail, since the same year marks the boom of the dot-com, the companies riding the rise of the early Internet. So, while his peers dream of becoming millionaires by building websites, he represents the United States in a contest where substance is what truly matters. A contrast that already says everything about the future ‘father’ of Claude.
Note
*1 Wikipedia, “Dario Amodei”;
2001-2016: from neuron to algorithm.
After graduating from high school, Dario Amodei begins his university studies at the renowned ‘Caltech’, the California Institute of Technology, where he is granted access to ‘Physics 11’: an experimental program reserved for the most talented students (*1). It’s during this period that he discovers the writings of Ray Kurzweil on the exponential growth of technology, where the author sets out a theory that makes a strong impression on the young student according to which the progress of computational power would inevitably lead to intelligent machines. He then transfers to Stanford, where he earns a bachelor’s degree in physics in 2006, and then to Princeton, where he undertakes a PhD in biophysics (*2).
Read more
His focus turns in particular to the way brain cells communicate and synchronize with one another, giving rise to complex behaviors. But biology alone is not enough: studying neurons ‘from the outside’ does not allow him to truly understand how intelligence works. This gradually draws him toward artificial systems.
After a postdoctoral position at Stanford School of Medicine, in 2014 he takes a position at Baidu (*3), a Chinese technology multinational, where he works on speech recognition based on deep learning. It’s in this context that he develops a fundamental insight that will shape his future: adding data, parameters, and computing power to neural networks makes them more capable (*4). The same observation is confirmed during his time at Google Brain (*5), marking the first stirrings of a method that would redefine the entire AI sector.
Notes:
*1 AI Wiki, “Dario Amodei”;
*2 PhD earned in 2011 (Hertz Thesis Prize). In his thesis dedication, Dario explicitly thanks his sister Daniela “for reminding him that the ultimate purpose of scientific work is to make the world a better place”;
*3: Baidu: Chinese technology multinational;
*4: Neural Networks: computational models inspired by the functioning of the human brain;
*5: Google Brain: the division of Google / Alphabet specializing in Artificial Intelligence and Deep Learning research..
2001/16
2016-2021: Dario Amodei’s years at OpenAI.
In 2016, Dario Amodei joins OpenAI: the organization co-founded, among others, by Sam Altman and Elon Musk, with the ambition of building Artificial Intelligence for the benefit of humanity.
He rises quickly through the ranks to become ‘VP of Research’, leading the development of GPT-2 and GPT-3, the language models that would revolutionize the entire field. In this context, he contributes to the creation of Reinforcement Learning from Human Feedback (RLHF): the technique that trains AI systems to behave in a helpful and responsible way through human feedback (*1).
Read more
These are intense years, during which he becomes increasingly convinced that safety cannot be a secondary element of AI progress, but must be its core principle. A conviction that, over time, comes into growing conflict with the company’s strategic decisions.
In 2021, he leaves OpenAI along with his sister Daniela and other colleagues (*2). Looking back on this decision, he would later say: ‘If you have a vision for how to do it, go and make it happen with the people you trust’ (*3).
Notes:
*1: darioamodei.com; Wikipedia, “Dario Amodei”;
*2: Wikipedia, “Dario Amodei”;
*3: PParaphrased from the original: Kitrum, “How Dario Amodei Brings His Vision for Safe AI to Life at Anthropic”, “How Dario Amodei Brings His Vision for Safe AI to Life at Anthropic”;
2016/21
The paper that started a field of research.
In 2016, the same year he begins working for OpenAI, Dario Amodei co-authors the paper ‘Concrete Problems in AI Safety’, one of the first to methodically address the risks inherent in Artificial Intelligence (*1). The academic work identifies five categories of concrete risk, including the ability of a system to circumvent its own instructions and to cause unintended side effects, proposing a specific research agenda for each.
Considered still today a reference point for the entire field, this content reveals how deeply this subject has always been at the heart of Amodei’s thinking.
Note
*1: Amodei et al., “Concrete Problems in AI Safety”, arXiv, 2016;
The Amodei siblings’bet: Anthropic.
In 2021, the Amodei siblings co-found their own company with former OpenAI colleagues, naming it ‘Anthropic’ (*1). This is no ordinary technology startup: it’s established as a Public Benefit Corporation, meaning it is legally required to balance shareholder returns with the public good.
But for Dario, this is not enough: he sets up the ‘Long Term Benefit Trust’, an independent body made up of five experts who hold no equity in the company, with the power to intervene in the composition of the board of directors should profit ever be placed above safety (*2). A mechanism of this kind, built into the very architecture of the organization, is yet further proof that for this digital entrepreneur, protecting AI reliability is not mere rhetoric but the foundation of everything he does.
2021
Notes
*1 Wikipedia, “Dario Amodei”;
*2 Wikipedia, “Anthropic”;
When Amodei turns down the chance to lead OpenAI.
In November 2023, the board of directors of OpenAI unexpectedly dismisses the company’s co-founder and CEO, Sam Altman. In the chaos that follows, Dario Amodei is approached with a two-part proposal: step in for the disgraced executive and consider a merger with Anthropic (*1). Amodei, seemingly without a moment’s hesitation, turns down both options. Altman, perhaps not by chance, is reinstated just a few days later. The episode is telling, revealing much about one of the men entrusted with shaping the future of Artificial Intelligence.
Note
*1 Wikipedia, “Dario Amodei”;
Dario Amodei’s essays.
Dario Amodei is one of the few leaders in the Artificial Intelligence sector to regularly commit to paper his thoughts on the future. In this regard, two essays in particular stand out for their depth.
The first is ‘Machines of Loving Grace’ (October 2024), in which he outlines an optimistic view of the new technology as a tool capable of dramatically accelerating the pace of scientific progress, ultimately improving the wellbeing of people (*1).
The second is ‘The Adolescence of Technology’ (January 2026), in which he identifies what he considers the five major risks of advanced AI (*2): misalignment with human values, misuse for destructive purposes, concentration of power, widespread job losses, and unexpected effects arising from its deployment.
Read more
These writings reflect opposing perspectives, one optimistic and one critical, both absolutely necessary for anyone who wishes to understand, with intellectual honesty, the impact of the digital revolution on humanity.
Notes:
*1: Dario Amodei, “Machines of Loving Grace”;
*2: Dario Amodei, “The Adolescence of Technology”;
Dario Amodei: A ‘Constitution’ for AI.
In December 2022, Anthropic publishes a paper with a certainly illuminating title: ‘Constitutional AI’: Harmlessness from AI Feedback’ (*1).
The idea behind it is as simple as it is revolutionary: to manage the behavior of new AI-based systems, rather than relying solely on the feedback of human evaluators (Reinforcement Learning from Human Feedback, RLHF), the intention is to provide a set of explicit principles directly to the systems themselves (RLAIF, or Reinforcement Learning from AI Feedback). Principles resembling a kind of ‘constitution’, inspired, among other things, by the Universal Declaration of Human Rights.
The ultimate goal is for models to learn by themselves to assess their responses against these criteria, independently identifying and correcting any alignment errors.
It’s no coincidence that this innovative, ‘values-based’ approach would soon after be applied to Claude, the LLM developed by Anthropic to compete with ChatGPT not only on a technical level, but on an ethical one.
Note
*1 Anthropic, “Constitutional AI: Harmlessness from AI Feedback”, December 2022;
Claude: AI ‘the Amodei way’.
In March 2023, Anthropic launches Claude: a large language model designed to compete directly with OpenAI’s ChatGPT (*1).
The challenge, however, goes beyond technology: its creator, Dario Amodei, aims to differentiate the product on a deeper level, that of values. While the AI assistant market focuses primarily on capability and speed, this LLM sets out to be above all a trustworthy interlocutor, capable among other things of reasoning with nuance, acknowledging its own limits, and declining potentially harmful requests. All of this through an innovative training method based on explicit principles.
The name itself is no accident: deliberately human, almost familiar, crafted to convey a sense of trust that goes beyond the system’s technical capabilities.
Read more
Within a few years, Claude establishes itself as one of the most widely used AI systems in the world, integrated into professional, academic, and creative contexts (*3). As a result, Anthropic’s valuation reaches 380 billion dollars in February 2026 (*4): concrete proof that safety and commercial competitiveness can coexist. As Amodei has always believed.
Notes:
*1: Anthropic;
*2: Anthropic;
*3: Wikipedia, “Anthropic”;
The origins of the name ‘Claude’.
Few people know that the name ‘Claude’ was not chosen by Dario Amodei solely for the sense of ‘familiarity’ he intended to convey to his ‘digital creation’, but also as a tribute to Claude Elwood Shannon (1916-2001), an American mathematician and engineer considered the father of ‘information theory’ (*1). In his celebrated 1948 paper, ‘A Mathematical Theory of Communication’, the scientist laid the mathematical foundations of all modern digital communication: foundations upon which the development of computers as we know them today, the internet, and ultimately artificial intelligence would be built.
Read more
An intellectual debt that Anthropic’s founder chose to make visible, but with discretion.
Note:
*1 Arimetrics, “What is Claude”, Taskade, “Anthropic History 2026”;
Dario Amodei says ‘no’ to the Department of War.
In February 2026, the US Department of War, under Secretary Pete Hegseth, requests a modification to the contract linking the government institution to the company. In particular, it demands the removal of the ban preventing the use of Claude in two specific cases: mass surveillance of the civilian population and autonomous weapons systems (*1).
Not surprisingly, Dario Amodei, consistent with his principles, flatly refuses the request. Hegseth responds by designating Anthropic as a ‘supply-chain risk’: an extraordinary measure, never before applied to an American company, that effectively renders it a kind of ‘pariah’ for federal agencies (*2). The move triggers obvious legal repercussions and a statement from Dario Amodei who, while reaffirming his patriotism, underlines the existence of ‘red lines’ that must not be crossed.
Notes
*1: Wikipedia, “Dario Amodei”;
*2: Wikipedia, “Dario Amodei”;
2001/16
Dario Amodei and the ‘weight of the future’.
The recent dispute between Dario Amodei and the US Department of War reveals, perhaps more than any other episode, the true nature of this man. Certainly not an entrepreneur chasing growth at any cost, but rather a scientist who constantly asks himself how to ensure that Artificial Intelligence does good, before it is too late.
His abilities, combined with his vision, have earned him considerable success, including in terms of public recognition and following.
It’s no coincidence that TIME included him among the hundred most influential people in the world in both 2025 and 2026 (*1).
Note
*1: TIME100 2026;
2016/21
Copyright information.
The images on this page were created using generative Artificial Intelligence tools, with the following exceptions:
Imagini Creative Commons:
- Dario Amodei, image owner TechCrunch (Wikipedia Link);
- Claude Elwood Shannon, image owner Tekniska Museet (Wikipedia Link);