Altman says OpenAI and Anthropic differ on AI regulation

Altman described OpenAI and Anthropic as having fundamentally different worldviews on AI regulation.[2] He also sought to distinguish his policy position from Anthropic’s, making that contrast part of the interview’s message.[1] The available account does not specify Anthropic’s detailed policy proposals, so the distinction should not be read as a full comparison of the companies’ regulatory plans.[2][1] For professionals tracking AI policy, the remarks offer a signal of disagreement between the companies, not a complete map of where their proposals diverge.

Altman’s argument is that AI’s benefits justify accepting some risks.[1] He said the world should accept some harmful consequences in exchange for those benefits.[1] The Guardian described his approach as light-touch AI regulation, while The Verge reported him saying some bad things will happen but AI is worth it overall.[3][4] These accounts frame his position as a willingness to tolerate some harms, rather than a claim that AI has no downsides.[1][4] The evidence does not quantify the risks or set out how they should be weighed.

Altman made the remarks in an exclusive interview with POLITICO’s Decoded, which POLITICO published on October 4, 2026.[1] POLITICO said the interview offered insight into how OpenAI is confronting political pressures.[5] The outlet also reported that OpenAI faces intensified competition with Anthropic.[5] Those details place the regulatory disagreement in a context of political pressure and competition, but they do not establish that either factor caused Altman’s position.[5]

Separately, OpenAI’s blueprint proposes a durable federal framework for governing increasingly capable AI systems.[6] The proposal includes strengthening CAISI as the federal government’s primary frontier-AI safety institution.[6] OpenAI also says the blueprint would mobilize a government-wide resilience plan for frontier AI’s national-security and public-safety challenges.[6] The company cited California’s SB 53, New York’s RAISE Act and Illinois’s SB 315 as state approaches to frontier-AI governance.[6] The blueprint’s proposals provide policy context, but the evidence does not say they were detailed in the interview.