The burgeoning efforts to foster America’s development of artificial intelligence, including for military use, largely overlook how our current advantage depends on immigrants. Without immigration reforms, this country’s days as the world’s AI leader may be numbered.

Foreign-born talent fuels the U.S. AI sector at every level. Immigrants lead many of America’s top AI companies, contribute groundbreaking original research in machine learning and other emerging disciplines, and handle much of the essential, ongoing work to deploy and manage AI technologies. Immigrants comprise two-thirds of U.S. graduate students in AI-relevant fields and founded many of America’s most successful companies active in AI, including Google, Tesla, and chipmaker Nvidia.

America’s world-class universities, leading companies, and high quality of life continue to attract the world’s best and brightest in AI, but an inflexible and restrictive skilled immigration system increasingly stands in their way. Many of the biggest problems have existed for years, if not decades. For example, annual limits on employment-based green cards haven’t budged since 1990, even as the national economy has doubled in size. Other problems are newer. The White House has voiced support for “merit-based” immigration, but in practice, stringent new adjudication policies have made applying for skilled immigrant status a longer, costlier, and more uncertain process. Processing delays are getting worse, prompting bipartisan concern in Congress and contributing to application backlogs in the millions.

These problems are undermining America’s AI sector. Renowned machine learning scientist Ian Goodfellow recently said visa restrictions on his foreign collaborators have been “one of the largest bottlenecks to our collective research productivity over the past few years.” Oren Etzioni, CEO of the Allen Institute for Artificial Intelligence, warns that “restrictive immigration policies are systematically depriving [American] universities of some of the world’s top talent.” MIT president Rafael Reif described similar trends as part of a “loud signal…that the U.S. is closing the door – that we no longer seek to be a magnet for the world’s most driven and creative individuals…[W]e should expect it to have serious long-term costs for the nation and for MIT.”

In the past, America might have been able to get by with flawed immigration laws: no matter the obstacles, AI talent would flock to the world’s unquestioned leader in science and innovation. But times are changing. Already, many of the world’s highest-valued AI startups are based outside the United States, and labs from Beijing to Bangalore are pursuing cutting-edge AI research. Meanwhile, other nations are overhauling their own immigration laws and offering generous incentives to immigrants with AI skills. In this competitive environment, status quo policy isn’t a viable option.

To maintain its advantage, the United States needs an efficient, accessible immigration system that welcomes the world’s AI talent. And that talent needs a clear and realistic path toward long-term legal status, since most who make the move want to remain in the United States. That means lifting outdated limits, overhauling restrictive regulations and agency policies, and making clear to the world – in rhetoric and in action – that America is open to talented AI workers, students, and entrepreneurs, no matter their origin. 

Achieving these reforms will also require addressing two common concerns: 

First, immigration is unlikely to crowd native-born U.S. citizens out of the AI sector. In fact, AI employers are hungry for talent and are hiring all the qualified candidates they can find. State and federal policies to support U.S.-born AI talent are welcome—but they’ll take years or decades to bear fruit, and won’t meet the market’s huge demand on their own. And leaving immediate supply and demand aside, America will always have an interest in ensuring that talented individuals come to live, work, and fuel AI innovation in the United States.

Second, the risks of theft or espionage don’t justify closing the country’s doors to foreign AI talent. To be sure, illicit technology transfer is a significant threat to the United States, and foreign visitors are sometimes to blame. But it’s unclear how well recent “extreme vetting” and screening policies can ferret out the small number of bad actors coming in from abroad. At the same time, these practices are making the immigration process slower, costlier, and less predictable, spreading fear and disillusionment among foreign AI students, workers and entrepreneurs. In turn, they make the United States less able to recruit the talent it needs to maintain leadership in artificial intelligence. That’s a serious national security threat in itself.

Immigration reform of any sort may be a tall order nowadays, but the dawn of the AI age is reason enough to redouble those efforts. Anyone interested in securing America’s advantage in artificial intelligence needs to work to open doors to talent from around the world. Otherwise, counterproductive immigration laws may make this country a second-rate AI power before long.