OpenAI recently released a policy proposal titled "Democratic Governance of Frontier AI: A blueprint for a federal framework," which advocates for civilian oversight of advanced artificial intelligence development. This stance differs from a newly issued executive order by the Trump administration, which designates the National Security Agency (NSA) as the primary body responsible for evaluating risks associated with frontier AI models. OpenAI suggests that regulatory responsibilities should instead reside with the Center for AI Standards and Innovation (CAISI), a division under the Commerce Department’s National Institute of Standards and Technology.

Despite the regulatory disagreement, OpenAI Chief Executive Sam Altman expressed public support for the executive order on the social media platform X. Altman is currently engaging in a week of scheduled discussions with White House officials and lawmakers, including meetings with Sen. Bernie Sanders, Speaker Mike Johnson (R-Louisiana), and Senate Minority Leader Chuck Schumer (D-New York). OpenAI executive Chris Lehane indicated that while the company is generally pleased with the administration’s order, it intends to advocate for transferring model evaluation authority to CAISI. OpenAI argues that CAISI possesses the sophisticated testing capabilities required for such oversight and notes that the company and its competitor Anthropic PBC have already established a close, collaborative relationship with the agency.

OpenAI also raised concerns regarding the executive order's plan for a "benchmarking" process to assess the cyber capabilities of frontier models. Lehane told Politico that the proposed framework may create ambiguity regarding the specific criteria used to designate a model as a "covered frontier model" and how much NSA scrutiny these systems will face. The executive order defines this benchmarking as an assessment of advanced cyber capabilities to determine whether a model crosses the threshold for enhanced regulatory oversight. Establishing clear, technical definitions for these capability thresholds remains a primary focus for upcoming discussions between the company and government representatives.

OpenAI recently released a policy proposal titled "Democratic Governance of Frontier AI: A blueprint for a federal framework," which advocates for civilian oversight of advanced artificial intelligence development. This stance differs from a newly issued executive order by the Trump administration, which designates the National Security Agency (NSA) as the primary body responsible for evaluating risks associated with frontier AI models. OpenAI suggests that regulatory responsibilities should instead reside with the Center for AI Standards and Innovation (CAISI), a division under the Commerce Department’s National Institute of Standards and Technology.

Despite the regulatory disagreement, OpenAI Chief Executive Sam Altman expressed public support for the executive order on the social media platform X. Altman is currently engaging in a week of scheduled discussions with White House officials and lawmakers, including meetings with Sen. Bernie Sanders, Speaker Mike Johnson (R-Louisiana), and Senate Minority Leader Chuck Schumer (D-New York). OpenAI executive Chris Lehane indicated that while the company is generally pleased with the administration’s order, it intends to advocate for transferring model evaluation authority to CAISI. OpenAI argues that CAISI possesses the sophisticated testing capabilities required for such oversight and notes that the company and its competitor Anthropic PBC have already established a close, collaborative relationship with the agency.

OpenAI also raised concerns regarding the executive order's plan for a "benchmarking" process to assess the cyber capabilities of frontier models. Lehane told Politico that the proposed framework may create ambiguity regarding the specific criteria used to designate a model as a "covered frontier model" and how much NSA scrutiny these systems will face. The executive order defines this benchmarking as an assessment of advanced cyber capabilities to determine whether a model crosses the threshold for enhanced regulatory oversight. Establishing clear, technical definitions for these capability thresholds remains a primary focus for upcoming discussions between the company and government representatives.