OpenAI Calls for Mandatory National AI Safety Regulation

OpenAI intends to collaborate with Congress to establish mandatory, capability-based national AI safety regulations. [1] While waiting for federal action, the company announced its support for four California bills designed to strengthen the AI safety ecosystem. [1] These state-level bills include SB 813 for independent safety assessments, AB 1405 regarding AI-auditor standards, SB 1119 for youth protections, and AB 1864 targeting AI-enabled biological threats. [1]

The organization plans to collaborate with other frontier labs to build voluntary industry standards, regardless of government support. [1] On a global scale, OpenAI will advocate for international approaches to measuring capabilities and determining when the advancement of model capabilities should be slowed or stopped. [1] OpenAI's Chief Scientist Jakub Pachocki recently stated that the rapid rise of machine intelligence calls for extreme caution. [1] The company noted that technical work inside individual laboratories will not be enough to handle these challenges. [1]

OpenAI is shifting its posture to explicitly invite legislative constraints, framing internal safety protocols as inadequate for frontier models. By simultaneously backing California's legislative package and pushing for federal mandates, the company is attempting to shape the regulatory landscape from the state level up. The call for international agreements on when to halt model development underscores the severity of the risks outlined by its leadership. What the sources don't address: How OpenAI defines the specific capability thresholds that would trigger its proposed mandatory federal safety regulations.