More than 1,100 artificial intelligence researchers, engineers and senior executives from leading AI companies have signed an open letter urging governments to support international efforts aimed at developing technical and governance mechanisms that could deliberately slow the pace of frontier AI development if required.
The signatories include employees and leaders from organizations such as OpenAI, Anthropic, Google DeepMind and Meta. Among the prominent names backing the initiative are Anthropic co-founders Jack Clark and Jared Kaplan, OpenAI Chief Scientist Jakub Pachocki, Meta Chief Scientist Shengjia Zhao and Google DeepMind's AI safety leader Anca Dragan. The initiative reflects growing concern within the AI community over the speed at which increasingly capable systems are being developed.
The letter calls on the US government to work with international partners to develop both technical and policy frameworks that would allow frontier AI progress to be deliberately paced if future systems begin advancing beyond human oversight. According to the signatories, the world currently lacks effective governance tools capable of managing rapid advances in AI capabilities while balancing innovation and public safety.
A central focus of the letter is automated AI development, often referred to as recursive self-improvement. The concept describes AI systems that can increasingly improve or develop their own capabilities with minimal human intervention. The signatories argue that while it is difficult to predict when such capabilities may emerge, there is a possibility that AI development could accelerate faster than society's ability to understand, evaluate or control the resulting systems.
The proposal does not call for an immediate halt to AI research. Instead, it advocates the creation of coordinated governance mechanisms that could be activated if frontier AI systems begin posing significant societal or security risks. The letter also encourages AI companies to voluntarily begin designing such coordination mechanisms even before governments introduce formal regulatory frameworks.
The appeal comes as discussions around AI governance gain momentum globally. In recent months, several industry leaders, including OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei and Google DeepMind CEO Demis Hassabis, have publicly argued for stronger oversight of the most advanced AI models, although they differ on how regulation should be implemented. Proposals have ranged from independent review agencies to phased pre-release evaluations for frontier AI systems.
The debate has intensified following recent incidents involving advanced AI models and growing concerns over cybersecurity, autonomous decision-making and the potential misuse of increasingly capable systems. At the same time, governments continue to weigh the need for safeguards against concerns that excessive regulation could slow innovation and weaken global competitiveness in the race to develop advanced AI technologies.
While policymakers in several countries are exploring AI governance models, comprehensive federal legislation in the United States has yet to materialize. In the absence of unified regulation, some state governments have introduced their own measures affecting AI infrastructure and deployment, highlighting the fragmented nature of current oversight.
The latest letter signals that concerns about AI safety are increasingly coming from within the industry itself. Rather than advocating restrictions on innovation, the signatories argue that coordinated international safeguards should evolve alongside frontier AI capabilities to ensure technological progress remains manageable, transparent and aligned with public interest.