OpenAI CEO Sam Altman Argues Society Must Accept Bounded Risks from Artificial Intelligence

OpenAI CEO Sam Altman argued that society must accept some "bad things" happening with artificial intelligence in order to unlock the broader benefits and human agency powered by the technology. Speaking on Politico’s Decoded podcast, Altman differentiated his regulatory philosophy from stricter peers, insisting that eradicating every minor abuse would require draconian restrictions that ultimately cause more harm than good.

Altman warned against setting impossible standards of zero misuse, such as completely preventing all hacks or scams, noting that the positive impacts of AI will outweigh negative outcomes by orders of magnitude. "Accept bounded risks, accept risks that we understand in exchange for the benefits, and liberty, agency, all those things," Altman said, while underscoring that developers must still draw a hard line against genuinely "catastrophic risks."

The debate over AI governance has intensified amid recent cybersecurity incidents involving frontier models, including unauthorized incursions on developer platform Hugging Face and Australian government systems, alongside four evaluation breaches acknowledged by Anthropic. Rival executives, including Anthropic CEO Dario Amodei, have advocated for far more stringent oversight, proposing an FAA-style federal agency empowered to audit, mandate testing, and halt unsafe model deployments.

While Congress has yet to pass comprehensive federal AI legislation, major tech firms continue to rely on a patchwork of voluntary commitments. OpenAI President Greg Brockman recently joined other industry executives at the White House to establish voluntary safety standards, even as OpenAI endorsed California legislation requiring public risk mitigation plans and safety frameworks.