Tom Brown: The Anthropic Engineer Who Conquered Washington

Le brief IA que les pros lisent chaque soir
Les 7 actus IA du jour, décryptées en 5 min. Gratuit.
Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.
Choisis ton rythme
Gratuit · Pas de spam · Désabonnement en 1 clic
Tom Brown: A Key Player in Anthropic's Negotiations
Tom Brown, one of the co-founders of Anthropic, has recently played a pivotal role in discussions with the U.S. government regarding the Fable 5 and Mythos 5 artificial intelligence models. Although his name may not be as well-known as some of his colleagues, Brown has been crucial in convincing the Department of Commerce to lift export restrictions on these technologies. This event highlighted the significance of his role not only in the company's technological development but also in its corporate diplomacy.
When tensions between Anthropic and the White House reached a critical point, it was not CEO Dario Amodei who took the reins of the negotiations, but rather Tom Brown. As the Chief Operating Officer, Brown successfully persuaded U.S. authorities to lift the restrictions, demonstrating his ability to navigate complex political waters. This episode revealed Brown's versatility, as he not only manages the company's computing power but has also established himself as an influential diplomat.
Michael Waxman, who co-founded a startup with Brown, expressed his confidence in him, stating that his presence in critical AI-related decisions is reassuring. Sarah Heck, head of public policy at Anthropic, also collaborated with Brown during discussions in Washington. A letter addressed to Brown by Lutnik regarding export controls further underscores his central role in these negotiations.
An Entrepreneurial Journey Before AI
Before dedicating himself to artificial intelligence, Tom Brown honed his skills in the world of Silicon Valley startups. A graduate of MIT, he first worked for Lingt, a language education startup, before joining MoPub, a mobile advertising company. In 2012, he co-founded Grouper, a group dating platform, with Michael Waxman as part of the Y Combinator accelerator program.
According to Waxman, Brown already possessed all the qualities of a leader at that time: integrity, honesty, and the ability to unite different disciplines within the company. After leaving Grouper, which struggled to compete with giants like Tinder, Brown decided to pivot towards AI. He embarked on an intense self-learning program, taking courses on Coursera, participating in projects on Kaggle, and studying reference materials in mathematics.
In 2015, Brown invited Waxman to join a study group on machine learning, an invitation Waxman declined, unaware at the time of the impact Brown would have on the field of AI.
From Grouper to OpenAI via Greg Brockman
Despite Grouper's failure against more popular dating apps, the company caught the attention of Greg Brockman, former CTO of Stripe. Impressed by Brown, Brockman often organized meetings for Grouper, allowing the two men to grow closer.
In 2015, when Brockman co-founded OpenAI, Brown immediately expressed interest in joining the team. He reached out to Brockman, stating his desire to contribute, even modestly, to this new venture. Brockman eventually offered Brown a position on a project related to the game StarCraft, although Brown did not initially work on machine learning.
During a discussion with the co-founders of Anthropic, Brown explained that the idea of contributing to AI safety attracted him to OpenAI, a field he had previously considered reserved for researchers.
Major Contributions in the AI Era
Although Tom Brown was not initially a researcher, he is now recognized for his significant contributions to artificial intelligence. His Google Scholar page lists over 140,000 citations, reflecting his impact in the field. In 2017, he co-authored a foundational paper with OpenAI and DeepMind on reinforcement learning from human feedback, a central technique in modern AI training.
After a stint at Google, Brown returned to OpenAI in 2018, where he played a key role in developing advanced language models. In 2020, he contributed to the publication of "Scaling Laws for Neural Language Models," which established development principles for increasingly large models. Later that year, he contributed to "Language Models are Few-Shot Learners," introducing the GPT-3 model, a major breakthrough in the industry.
Brown emphasized that the scaling and safety teams at OpenAI, led by Daniela and Dario Amodei, took the idea of "scaling laws" very seriously. These reflections led to the creation of Anthropic in 2021, with the ambition of ensuring that transformative AIs are developed responsibly.
A Leader in the Quest for Computing Power
Unlike other AI leaders who focus on research or product development, Tom Brown must negotiate with the largest players in the industry to secure the computing power necessary for Anthropic. Under his leadership, Anthropic's computing team has struck strategic deals with giants like Amazon, Nvidia, and Google, diversifying the sources of computing power.
In the spring, faced with growing demand for the Claude Code product, Brown orchestrated a multi-billion dollar deal with SpaceX to utilize their Colossus data center. He also represented Anthropic at Amazon Web Services' re:Invent conference and was cited by Google Cloud to promote collaboration on AI infrastructure.
In a discussion with his co-founders, Brown mentioned the challenges of convincing certain partners about the speed at which AI could transform the world. He admitted to having changed his own mind over time, moving from initial skepticism to a growing conviction.
A Philanthropic Commitment
Like his fellow co-founders at Anthropic, Tom Brown is committed to giving away the majority of his wealth. Bloomberg estimates his net worth at nearly $8 billion, thanks to his stake in Anthropic, valued at $965 billion. Brown has pledged to donate 80% of his wealth, a commitment shared by several of his co-founders.
The lifting of restrictions by the White House comes at a crucial time for Anthropic, which is considering an IPO. This move could provide welcome liquidity for employees and enhance the company's expansion capabilities.
The fact that Brown was able to step in for Amodei during negotiations with the U.S. government speaks to the depth of leadership within Anthropic. While Brown may have taken a different approach than his CEO, he effectively advocated for the company's interests. When asked by Business Insider for his career advice, Brown recommended surrounding oneself with inspiring people, emphasizing the importance of positive influence in personal and professional development.
Brief IA — L'actualité IA en français
L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.