As OpenAI and Anthropic race towards their IPOs, they’re seemingly getting more open about taking potshots at one another.
The latest instance comes from Sam Altman, who in a recent interview laid out what he sees as the two ways AI could go badly wrong, and in the process appeared to take a swipe at the philosophy driving Anthropic and its co-founder Dario Amodei. Altman was asked about his fears for how the technology could develop, and his answer split the risk into two buckets before landing somewhere that sounded a lot like a reference to Anthropic’s approach to safety and control.
“AI somehow just becomes too powerful in a way that we can’t guarantee the control we want,” Altman said. “And the other is power gets too centralised, where you have, you know, one company or model or person with too much power. And in both of these, the fundamental thing is, like, I think it’s a very anti-human position for either of these things to happen,” he added.
“The right approach is to say, like, we want people deeply in control of the future. We want people deeply empowered. People are the whole point of this all. Like, this is, we are not going to sit here and, you know, gradually hand over control to an AI model because we don’t trust or like people. And, you know, it’s like a very misanthropic thing to say we’re going to just put all of our trust in this model and let it have all the power and decision making over the world. But I think there are some people in the world who think that’s the right outcome.”
The framing here is hard to read as anything other than pointed. Anthropic has built its entire identity around the idea that AI risk needs careful management, and Amodei himself has floated scenarios where a small number of trusted, safety-focused actors end up with outsized influence over how the most powerful systems are built and deployed. Investor Gavin Baker recently alleged on the All-In podcast that Amodei had told people internally that Anthropic could end up being the only private company left in the world, with nothing standing beside it but governments. Amodei’s team denied this directly, calling the account false, but the claim spread fast and fed exactly the narrative Altman appears to be gesturing at here — that Anthropic’s worldview edges toward concentrating power and decision-making in the hands of a very small circle, AI included, rather than distributing it broadly among people.
Altman’s line about not wanting to “gradually hand over control to an AI model because we don’t trust or like people” also reads as a jab at the broader camp that Anthropic represents, a camp that has often argued caution and centralized oversight are necessary precisely because the alternative is too dangerous to risk. Altman is essentially arguing the opposite: that treating people as too unreliable to be trusted with power, even power mediated through AI, is itself a bad starting assumption.
There’s also a subtler echo running through Altman’s comments of Amodei’s own essay, Machines of Loving Grace, in which the Anthropic CEO sketched out an optimistic vision of a world transformed by powerful AI while spending considerable time on the risks of getting there wrong. Amodei’s essay leaned into the idea that a small set of responsible actors managing the transition carefully was the safer path forward. Altman’s “very anti-human” framing seems designed as a direct counterpoint to that idea, arguing that any vision built around ceding ground to AI systems or concentrating control, however well-intentioned, ends up working against the people it claims to protect.
Neither Altman nor OpenAI named Anthropic directly, and it’s entirely possible the comments were meant more broadly, aimed at anyone in the AI safety world who leans toward caution over distribution of power. But given the timing, with both companies gearing up for IPOs and the rivalry between them growing more public by the week, it’s difficult to read the remarks as anything other than aimed squarely across town.