Dario Amodei vs Sam Altman: A decade-long feud in the battle for AI's future

Dario Amodei and Sam Altman once built OpenAI together — and together determined the path the race for artificial intelligence would take. Today, the CEO of Anthropic openly calls his former boss and his company a threat — and his words grow harsher with each passing month. In private conversations with colleagues, Amodei's rhetoric has taken on unexpected sharpness.

He compared the legal dispute between Altman and Elon Musk to the confrontation between Hitler and Stalin — suggesting that one should observe this clash dispassionately, without sympathy for either side. When OpenAI president Greg Brockman directed $25 million to a pro-Trump political action super PAC, Amodei called it outright evil. In public speeches, he has repeatedly likened OpenAI and other competitors to tobacco companies — structures that know perfectly well the harm of their product but continue to sell it.

This is not merely competitive rhetoric. There is a personal history behind it. Amodei spent several years at OpenAI as vice president of research — and was one of the key architects of its technical strategy.

Together with his sister Daniela and a group of about ten colleagues, he left the company in 2021 — officially due to disagreements over safety approaches and the pace of commercialization. The Anthropic they founded was positioned from the start as an alternative: slower, more cautious, with a strict focus on aligning AI values with human interests. Over the years, this narrative has acquired an increasingly pronounced antagonist.

OpenAI under Altman's leadership has moved from a nonprofit research organization to a commercial entity with a valuation exceeding one hundred billion dollars. ChatGPT became the first truly mass-market AI product — more than 400 million monthly active users. The success of GPT-4 and subsequent models cemented the company's market leadership.

Against this backdrop, Anthropic with Claude and strict internal protocols increasingly looks more like the conscience of the industry than its technological vanguard. Amodei is clearly not prepared to accept this role. The comparison to tobacco companies is not a random metaphor but a deliberate narrative with a specific accusatory meaning.

The tobacco industry knew for decades that its product was carcinogenic and concealed this data from the public. Applying the same logic to AI, Amodei is essentially accusing competitors of deliberately accelerating the development of systems whose danger they understand just as well as Anthropic specialists. This is a serious accusation — and the industry perceives it as such.

The question of which of the two models of AI development is destined to prevail — the commercial or the safe one — remains open. But the fact that this conflict is increasingly influencing real decisions: who receives funding, what standards are adopted, how regulators view the industry — is no longer in doubt. Disputes over AI technology supplies to the Pentagon that flared up in recent months only laid bare the depth of the gap between the two companies in fundamental values.

While OpenAI actively takes on state defense contracts, Anthropic takes a far more cautious stance. Amodei has repeatedly stated publicly the need for serious reflection on how AI-based systems are integrated into military and intelligence structures — and why the speed of implementation should not outpace understanding of risks. Behind all this rhetoric lies the question that will define the industry in the coming years: who sets the pace of the race — the one betting on speed and scale, or the one insisting on a pause for safety?

The personal feud between Amodei and Altman has long transcended the bounds of corporate conflict. It is a public clash of two fundamentally different philosophies of technological development — and what AI enters every home depends on which one prevails.