U.S. President Trump signed an executive order on Thursday local time aimed at weakening state laws restricting the artificial intelligence industry, a major victory for technology companies that have opposed such regulations. The executive order gives the attorney general broad authority to sue states and overturn laws that are detrimental to "the United States' continued dominance in global AI," putting dozens of state laws related to AI safety and consumer protection at risk. Trump has directed federal regulators to withhold federal funding for projects like broadband if states refuse to repeal the laws.
Trump has said it is crucial for the United States to stay ahead in artificial intelligence and criticized state laws for creating a confusing "patchwork" of regulations. He said the order would create a unified federal regulatory framework that would supersede state laws.
“There has to be only one source,” Trump told reporters in the Oval Office, flanked by officials including David Sacks, the head of artificial intelligence and cryptocurrency. "You can't go to 50 different sources."
Trump has increasingly embraced the artificial intelligence industry in recent years, signing executive orders to limit regulation, provide access to federal data, and relax restrictions on companies building artificial intelligence infrastructure. He has also lifted barriers to the export of chips needed for artificial intelligence (including this week) and publicly praised tech company leaders. He also gave Sachs, who is also a Silicon Valley investor, huge influence on policymaking.
The executive order sparked widespread bipartisan opposition after it was announced Thursday. Legal experts say the order is likely to be taken to court by states and consumer groups, arguing that only Congress has the authority to overturn state laws.
“If Trump succeeds in weakening state laws, he should provide a strong national regulatory standard for artificial intelligence,” said Wes Hodges, acting director of the Center for Technology and Humanity, a subsidiary of the Heritage Foundation, a right-wing think tank. “Doing so without establishing corresponding national protections simply provides immunity to big tech companies.”
The new generation of generative artificial intelligence technology, which can imitate human writing and voice and generate lifelike videos and images, is developing rapidly. But the technology can also be misused to deceive consumers, and chatbots have been documented providing harmful advice to minors.
In the absence of federal oversight, states have enacted their own AI safety laws, requiring companies to take specific security measures and setting boundaries around the use of the technology.
According to the National Conference of State Legislatures (NCSL), all 50 states and territories have introduced relevant legislation this year, and 38 of them have passed about 100 laws.
California passed a law requiring the largest artificial intelligence models, including OpenAI’s ChatGPT and Google’s Gemini, to undergo security testing and disclose the results. South Dakota passed a law banning the use of lifelike artificial intelligence-generated videos, known as "deepfakes," in political ads just months before the election. Utah, Illinois and Nevada have passed laws related to AI chatbots and mental health, requiring notification that users are interacting with chatbots and imposing limits on data collection.
At the signing of the executive order, Sacks said the administration would target the most "onerous" state laws, citing some state-level legislation that could "embed ideological bias into AI models."
States are also passing a growing number of child safety regulations targeting AI chatbots and social media businesses that use AI technology in their services.
Thursday's executive order said it would not replace laws related to child safety but did not provide details.
AI companies have been lobbying fiercely in Congress and the White House to try to roll back state-level regulations. Earlier this year, some lawmakers tried to include a ten-year freeze on state AI laws in a domestic policy bill, but it stalled amid strong bipartisan opposition.