OpenAI Argues Case For AI-Friendly US Rules

ChatGPT developer OpenAI has proposed that US AI companies be granted relief from proposed state regulations if they voluntarily share information on their models with the federal government.

The proposal came in a 15-page proposal on US AI policy submitted as part of the Trump administration’s plans to create an AI Action Plan by July.

If AI companies work with the US federal government’s AI Safety Institute, they could be provided “with liability protections including pre-emption from state-based regulations that focus on frontier model security”, OpenAI said in the paper.

Deregulation

It described proposed state AI laws as “overly burdensome” and emphasised that AI companies’ federal participation would be “on a purely voluntary and optional basis”.

The company suggested measures that would help the federal government adopt AI more quickly, which it said could allow services to be accessed 12 months earlier than current processes.

Government use of AI is controversial because it would involve feeding government-held personal information on citizens into privately operated systems.

OpenAI also said the government could work with the private sector on AI models trained with highly sensitive information for national security purposes.

“The government needs models trained on classified datasets that are fine-tuned to be exceptional at national security tasks for which there is no commercial market – such as geospatial intelligence or classified nuclear tasks,” the company wrote.

OpenAI released ChatGPT Gov, built specifically for government use, in January.

The company said an “export control strategy” should be developed to promote global adoption of US AI systems.

Copyright restrictions

It argued US copyright strategy should ensure American AI models retain their “ability to learn from copyrighted material”.

OpenAI has been sued by numerous groups, including news and creative organisations, saying their copyrighted material was used without consent. OpenAI argues it works within fair-use principles.

The success of Chinese AI start-up DeepSeek shows the US lead in AI “is not wide and is narrowing”, OpenAI wrote.

If China’s developers have “unfettered access to data and American companies are left without fair use access, the race for AI is effectively over”, the company wrote.

Donald Trump in January revoked the previous administration’s executive order on AI safety and issued a new order stating that the country’s policy is to “sustain and enhance America’s global AI dominance”.