The Problem
The founding premise of Anthropic was a claim about *how* frontier AI should be built, not merely *whether* it should exist. In their published Core Views essay, the founders argued that if powerful AI systems were coming — a claim they treated as very likely — the safest path was for a subset of the most safety-conscious researchers to be at the frontier building them, rather than ceding that ground to organisations less focused on the risk surface.
Dario Amodei had run research at OpenAI for several years and was familiar with the frontier both technically and organisationally. Daniela Amodei ran operations. They left OpenAI in late 2020 with five other researchers to found Anthropic in 2021, structured as a public-benefit corporation with an explicit safety mission.
The problem was not building a large model. It was building an organisation that could sustain frontier-scale training runs while keeping safety at the centre — organisationally, not rhetorically. That combination had few precedents.
The Journey
Amodei's academic background is in physics — a Princeton PhD, per the Anthropic team page and coverage in Time's 100 Most Influential People in AI. His technical work spans machine learning research at Baidu and OpenAI, with authorship on multiple influential papers including "Concrete Problems in AI Safety."
Anthropic began operations in early 2021 and released its first commercial product, Claude, in early 2023. The company's fundraising trajectory has been reported in Bloomberg, The Wall Street Journal, and Reuters: successive rounds from Google, Amazon, Spark Capital, Lightspeed, and others, taking the company's valuation past $60 billion by mid-2025.
Responses
Join the conversation
You need to log in to read or write responses.
No responses yet. Be the first to share your thoughts!