OpenAI chief executive Sam Altman has drawn a clear line between his company’s regulatory philosophy and that of rival Anthropic, arguing that the broader benefits of artificial intelligence justify accepting a degree of risk and that the technology should remain widely accessible rather than restricted to a single powerful actor.
Speaking in an exclusive interview with Decoded, the new Politico tech publication, Altman acknowledged meaningful philosophical distance between the two companies. Anthropic, led by Dario Amodei, has consistently been the industry’s most vocal advocate for stringent AI safeguards, a position that until recently put it at odds with OpenAI’s preference for lighter-touch regulation. Altman’s view is that society should be prepared to accept some negative consequences in exchange for the transformative potential AI offers to ordinary people.
Altman was pointed in his criticism of the idea that advanced AI development should be concentrated within a single laboratory. He described such an arrangement as entirely unacceptable, arguing it conflicts with the principle that people should have genuine agency over a technology with enormous positive potential. While he acknowledged that harms including scams and misuse are an inevitable feature of any widely deployed platform, he argued that positive use would outweigh the negative by orders of magnitude.
That tolerance for risk has its limits, however. Altman was clear that he does not accept what he described as truly catastrophic outcomes, particularly any serious erosion of human control over AI systems.
The remarks come despite a notable narrowing of the gap between the two companies on policy. In recent months, OpenAI has shifted its position considerably, endorsing stricter state-level safety legislation and backing a bipartisan Congressional proposal that would require leading AI developers to bring in independent safety evaluators. Altman has also aligned with Amodei’s recent call to moderate the pace of development on the most powerful AI models.
The Decoded interview ranged more broadly across the AI landscape, touching on President Donald Trump’s new AI accord with major technology companies, a wave of autonomous hacking incidents attributed to OpenAI models, questions around the company’s potential legal liability for those events and Altman’s views on the flow of AI industry money into US politics.
