U.S. officials seek easier pathways to ban AI exports and keep tech secrets in

Amid the global race to obtain high-level AI models, legislators in the U.S. have unveiled a new bill that would make it easier to ban exports of AI tech to safeguard against mal-intended foreign actors.

Legislators in the United States have unveiled a bipartisan bill that would aid the Biden administration in imposing export controls on top artificial intelligence (AI) models created in the country. 

In the late evening of May 8, House Republicans Michael McCaul and John Molenaar and Democrats Raja Krishnamoorthi and Susan Wild announced the bill, which aims to make any future regulations on AI exports bypass legal challenges.

It would also give the Commerce Department “express authority” to hinder American citizens from employment with foreign agents in the development of AI systems that would pose a potential risk to U.S. national security.

Under the current legislation, it is more difficult for the U.S. Commerce Department, which oversees export policy, to regulate open-source AI models. If the newly suggested bill passes it would minimize any roadblocks to regulating open source AI exports.

This move follows a Wednesday report from Reuters that said the U.S. is ready to take action to safeguard its homegrown AI from China and Russia. It said it will implement export controls on its most advanced proprietary AI models.

China has been known to utilize open-source models originating from the U.S., like Meta’s Llama models, to develop its own AI.

Related: AI-powered smart contracts could be ‘transformative’ — Ava Labs founder

It also follows a recent agreement on April 16 between Microsoft and the United Arab Emirates (UAE) AI company based in Dubai, the G42. Microsoft pledged a $1.5 billion deal with the UAE to bolster local AI innovation.

This deal will also give the UAE access to cutting-edge AI technologies developed in the U.S. by Microsoft.

It is one of the many deals made by Microsoft with foreign governments regarding the development and use of its AI models and cloud services.

However, the U.S.’s main concern continues to be high-level AI technology falling into the hands of the Chinese government. In November 2023, 01.AI – one of China’s most high-profile AI companies founded by a former Google executive Lee Kai-fu, revealed that its AI model Yi-34B was built using Meta’s Llama system.

This is one of many steps the U.S. government has taken to sanction China and minimize its reach into U.S.-made AI tech. In January, the Chinese AI firm Baidu faced potential sanctions over the People’s Liberation Army’s (PLA’s) alleged use of the company’s technology for military endeavors.