A proposal unveiled by a bipartisan group in the US House of Representatives on Wednesday aims to make it easier for the Biden administration to impose export controls on AI models, in order to safeguard American technology from malicious foreign actors.
The bill, introduced by Republican representatives Michael McCaul and John Molenaar, along with Democratic representatives Raja Krishnamoorthi and Susan Wild, also explicitly authorizes the Department of Commerce to prohibit Americans from collaborating with foreigners on the development of artificial intelligence systems that pose a risk to US national security.
The specific process, at present, this bill has been submitted to the House Foreign Affairs Committee.
The specific provisions of this bill are an amendment to the "Export Control Reform Act of 2018" to prevent foreign adversaries from exploiting American artificial intelligence and other supportive technologies for nefarious purposes. This bill aims to shield future AI export regulations from legal challenges and has taken input from Biden administration officials during its drafting.
The reasons behind this move, as cited by the U.S. government and private sector researchers, stem from concerns that adversaries could utilize these models to launch aggressive cyber attacks or even develop potent biological weapons. These models sift through vast amounts of text and images to synthesize information and generate content. For instance:
AI-based deepfake technology, which swaps faces, can create realistic yet fabricated videos trained by AI algorithms on numerous online clips, blurring facts and fiction in the polarized world of U.S. politics.
AI-supported image creation tools provided by companies like OpenAI and Microsoft, as highlighted in a March report by researchers, despite each website having policies against creating misleading content, could be used to produce photos propagating false information related to elections or voting.
The U.S. Department of Homeland Security previously stated in its 2024 Homeland Threat Assessment report that malicious actors might employ AI to "develop new tools" for launching "larger-scale, faster, more efficient, and more evasive" cyber attacks on critical infrastructure, including pipelines and railways.
Amid these concerns, a bipartisan legislative group in the U.S. House of Representatives has taken action to draft the aforementioned bill.
However, as of now, the U.S. Department of Commerce and the White House have not commented or responded to the proposal.
In reality, with the release of open-source models like Meta's Llama series, developers and businesses worldwide are leveraging this technology. Banning large AI models could not only severely impact the practical applications of AI models but also erode confidence in open-source gradually.
According to reports, under current U.S. laws, regulating the export of open-source AI models is much more challenging for the U.S. Department of Commerce, which oversees U.S. export policies, as open-source AI models can be freely downloaded globally.
If this bill is approved, it will eliminate barriers in the International Emergency Economic Powers Act that regulate the export of open-source artificial intelligence and grant the Department of Commerce explicit authority to regulate AI systems.
Following the proposal, during a routine press briefing on May 10th, a journalist asked about this issue. Chinese Foreign Ministry spokesperson Lin Jian stated that the facts prove that mutually beneficial economic and trade cooperation between China and the U.S. benefits both countries and their people. Politicizing, weaponizing, and ideologizing economic and technological issues, forcefully pushing for decoupling, disrupts normal trade, investment, and supply chain stability globally and between the two countries, which is not in the interest of any party, including the U.S.
Lin Jian emphasized that the U.S. should honor its commitment not to seek decoupling from China, not to hinder China's development, cease protectionist practices, stop technological blockades against China, and cease disrupting the international economic and trade order. China will take necessary measures to firmly defend its legitimate rights and interests.
