In remarks to Politico for the debut issue of its new Decoded newsletter and podcast, Sam Altman of OpenAI contended that society must tolerate certain downsides as the price of AI's advantages. The comments reveal persistent disagreement between OpenAI and Anthropic on the path forward for industry oversight.
Altman made clear his position on the trade-offs inherent in deploying powerful AI systems. "We believe that the world should accept some bad things happening for the benefits of this technology and people having the agency," Altman told Politico. He rejected the notion of sacrificing progress to eliminate hacking, fraud and misapplication of the technology. Rather, Altman expressed confidence that AI will generate substantially more positive outcomes than negative ones. However, he drew a line at accepting existential threats, particularly scenarios involving a loss of human control over AI systems.
The disagreement with Anthropic runs deeper than Altman's risk calculus. When asked about the competing vision—where a single laboratory would control advanced AI and distribute its benefits—Altman rejected it outright. He characterized such an arrangement as "a completely unacceptable trade-off". Anthropic's spokesperson countered by clarifying that the company's regulatory framework targets only frontier models, the most cutting-edge AI systems.
The exchange follows Anthropic chief executive Dario Amodei's recent call for the sector to decelerate its pace of model development. Altman endorsed Amodei's position within days of its publication three weeks prior. OpenAI announced it would grant external testers equivalent access to its systems as its own staff members receive.
Altman's interview with Politico also touched on the AI safety commitment signed by industry leaders at the White House, as well as the administration's push to rebrand artificial intelligence as "superintelligence." Meanwhile, OpenAI is among four AI firms scheduled to provide sworn testimony at a New York City Council hearing focused on AI safety, set for Monday.
Source: The Next Web
