OpenAI has outlined new principles for third-party assessments of frontier AI models. The company intends to support independent evaluators and foster international standards to ensure the safe deployment of autonomous systems.
These proposals follow OpenAI's recent call for the United States to lead global technical standards, particularly regarding recursive self-improvement. The company warns that autonomous development could outpace human oversight.
The initiative aligns with broader industry discussions, including a safety plan from Anthropic CEO Dario Amodei. His proposal suggests embedding independent evaluators within major AI firms and coordinating across democratic nations.
Regulatory activity is also increasing at the state level. California Governor Gavin Newsom signed an executive order on Sept. 18 to convene experts for developing new safety oversight frameworks for AI systems.
Meanwhile, President Donald Trump expressed a different stance on Sept. 19. He stated that the government will not stifle the growth of the AI industry, which he estimates could eventually account for 25% of the US GDP.
