Altman Says AI’s Upside Is Worth Accepting the Damage Along the Way
Altman Says AI’s Upside Is Worth Accepting the Damage Along the Way
Sam Altman has drawn a sharper line between OpenAI and its safety-minded rival Anthropic, arguing that society should accept some of AI’s damage rather than slow the technology’s advance.
In a Politico interview, Altman framed the disagreement as fundamental: OpenAI is prepared to live with “bad things happening” in pursuit of the technology’s benefits, a formulation that critics note centers the promise of AI rather than a clearly defined benefit to humanity.
The OpenAI chief later put more detail on what that bargain could involve. Speaking on Politico’s Decoded podcast, Altman said the world should accept hacks, scams and other harms because people will use AI to do “tremendously orders of magnitude more good stuff.”
The timing matters. OpenAI has been pushing for a lighter-touch regulatory approach even as concern grows over increasingly capable AI systems — including the company’s struggles to prevent agents from hacking real-world targets. Altman did not argue for a regulatory free-for-all, however. He described OpenAI as a “pragmatic centrist” position between Anthropic’s more cautious posture and those who would impose little or no oversight.
His warning runs in two directions. Restrict AI too heavily in the name of safety, Altman said, and power could end up concentrated in a handful of companies, individuals or perhaps a single country — another route by which things could “go horribly wrong.” Yet he also acknowledged the far darker possibility that humans could lose control of AI systems.
That leaves the central divide exposed: Anthropic’s approach treats the prevention of foreseeable harms as a reason to move carefully; Altman’s treats those harms as risks society may have to absorb. Both camps see danger in AI. They disagree over whether the greater danger lies in moving too fast or holding back.
Write a comment