Altman Calls for AI Pacing to Prevent Loss of Control

Leaders at Anthropic and OpenAI have jointly called for a coordinated approach to pace the development of frontier artificial intelligence, citing escalating safety concerns and the need to maintain global competitiveness. The proposal emerges amid growing industry awareness that AI systems are advancing at an unprecedented speed, primarily driven by their increasing capability to autonomously enhance subsequent generations. Dario Amodei, chief executive of Anthropic, outlined the urgency of the situation in a recent policy essay, highlighting the risk of recursive self-improvement, a mechanism where AI models continuously optimize their own architectures. Amodei warned that without careful oversight, such self-directed progress could outpace human comprehension and control. To mitigate these risks while preserving commercial advantage, he proposed a three-step regulatory framework. The initial measure requires leading AI laboratories to grant external evaluators employee-like access to their internal systems and development pipelines. Anthropic has already committed to implementing this transparency measure immediately. The subsequent steps involve establishing unified safety standards across the industry, voluntarily restricting the velocity of unchecked capability gains, and pursuing global coordination mechanisms to prevent fragmented or reckless development. Sam Altman, chief executive of OpenAI, publicly aligned with Amodei’s position, confirming that OpenAI would adopt the independent evaluator framework. In a series of statements, Altman emphasized that industry pacing is distinct from halting progress. He stressed that rapid innovation must continue but must be carefully balanced with robust alignment research and advanced monitoring capabilities. According to Altman, no amount of competitive pressure should justify safety compromises that allow system capabilities to exceed oversight capacities. The collaborative stance between two of the sector’s most influential firms signals a potential shift toward industry-led self-regulation. By prioritizing standardized safety protocols and external auditing, the AI community aims to preempt more stringent government intervention while maintaining technological leadership. Both executives acknowledged that domestic industry coordination must serve as a foundation before scaling safety frameworks internationally. Governmental bodies are expected to play a critical role in harmonizing cross-border regulations and ensuring equitable compliance. The proposed measures reflect a broader industry consensus that unchecked acceleration poses systemic risks to technological stability and public trust. By institutionalizing transparency, establishing common safety benchmarks, and committing to deliberate development cycles, leading AI labs are attempting to align innovation with responsible oversight. The success of this coordinated approach will likely determine the trajectory of frontier AI development, influencing both commercial deployment and regulatory policy in the coming years.