OpenAI has proposed the establishment of international safety standards for artificial intelligence. The company suggests that the US government should lead a global coalition to set these requirements, drawing on established frameworks from the aviation and financial sectors to ensure secure information exchange between nations.
The startup emphasizes the need for specific regulations concerning AI systems capable of recursive self-improvement. OpenAI argues that such autonomous systems should not be developed until their safety is fully verified, noting that current progress must remain under human control and align with universal values.
The initiative includes a call for secure communication channels between the US and China to report potential AI security threats. This proposal follows discussions between US Treasury Secretary Scott Bessent and Chinese Vice Premier He Lifeng, which confirmed a mutual interest in maintaining a dialogue regarding AI safety.
The proposal has received support from Google DeepMind CEO Demis Hassabis, who suggested creating a public-private regulatory body. Meanwhile, Anthropic continues to advocate for voluntary development slowdowns and the implementation of mandatory external safety audits for leading AI models across the industry.
Despite these proposals, the US administration remains cautious about increasing state intervention. President Donald Trump has expressed concerns that voluntary restrictions on AI development could jeopardize the country's competitive position against China, viewing such regulatory demands as potentially counterproductive.
