Congress Should Set One AI Standard Before It Is Too Late

Unfortunately, thousands of state bills regulating artificial intelligence (AI), often contradictory and even impossible to comply with, have been proposed across the country in the last eighteen months. Many have even been enacted. Congress has an immediate opportunity to make sure that the United States can continue its global leadership in AI by passing legislation establishing a national framework to prevent a patchwork of AI laws in the states that will harm our ability to lead on AI.

Artificial intelligence is changing how we create medicines, analyze our finances, build bridges, and even how we weed our crops. From medicine to the environment to transportation and more, AI will bring unprecedented benefits, and with them, unprecedented prosperity.

The early days of electric power in the 19th century were full of news stories about dangers from the technology. Imagine if states had engaged in broad and intrusive regulation of that industry in the months after Edison’s first practical lightbulb. Not enough was known yet about safety for meaningful regulation, but electrification could have been stopped dead in its tracks before it had even begun, the victim of unfounded fears. The damage to ordinary American families would have been extensive. It would have denied us the benefits of refrigerators, safe lighting, washing machines, and even well pumps at a time when few knew what those would mean for their lives.

Our current moment of progress, brought by AI, can only continue to move ahead smoothly if Congress creates a national framework to prevent thousands of state laws from stifling common sense and economic growth. Congress can take decisive action to preserve American leadership in AI in the next few weeks.

It is understandable that states want to legislate on AI. There are many legitimate fears, like deepfake driven scams. There are also many illegitimate ones; “The Terminator” was entertainment, not a documentary. However, states already have strong tools available to deal with the true problems we currently know about. Existing consumer protection, privacy, anti-discrimination, child protection, and product safety laws already apply to AI uses. These give state regulators wide latitude to address harms by enforcing laws that are already on the books.

Many state laws on AI anticipate harms that either have not occurred or are, at very best, far fetched. These speculative state laws also supersede Congress’s critical role in regulating interstate commerce. The founders granted Congress that power to avoid the patchworks of voluminous, contradictory laws with significant compliance costs that states are creating today.

If these state regulations are allowed to stand, the result will be an unstable, unpredictable environment for the development of new AI models and applications. It will be especially difficult for small AI startups to follow different laws across many states as the legal costs may exceed their abilities. Innovation, engineering, and speed will all be sacrificed to keep up with an ever-changing, anti-innovation legal landscape. AI companies of every size will struggle to comply with a costly and confusing AI regulatory regime, and only the largest may survive at all.

Congress has a chance to act now to avoid this mess. AI is an inherently national issue, and Congress has the authority to regulate technologies that operate across state lines.

Fortunately, must-pass legislation is now before Congress: the National Defense Authorization Act. Legislation to preempt state AI laws would be a wise addition to this year’s NDAA, especially because of the impact of state regulation on military preparedness. The negative impacts of state AI regulation will particularly harm our national defense. Our adversaries will welcome a wholesale slowdown in American AI development as they race ahead, seeking to outpace us. A morass of mutually contradictory state regulations will reduce AI innovation to a trickle and weaken our military at a time when China is pushing ahead full steam on AI research and development.

If state regulations are not preempted, the United States runs the risk of losing its edge in AI, which is essential to equipping our warfighters, driving our military industrial base, and protecting America from foreign threats. Without robust AI capabilities, the US could very well fall behind adversaries in military capabilities.

A national AI regulatory framework will not eliminate state authority in areas like consumer protection, child protection, privacy, or civil rights laws, or result in some imagined lawless space for AI development. It would simply prevent states from imposing conflicting AI-specific frameworks that bind the nation as a whole.

Preemption doesn’t have to be permanent, but it must pass Congress to keep our AI innovators innovating. Such legislation will give lawmakers the time and space needed to see and understand AI’s actual interaction with society, and eventually create laws based on wisdom, not Hollywood movie scenarios or fear.

There is ample precedent for such action. In the 1990s, faced with a similar morass of local legislation, Congress correctly decided that a patchwork of state Internet regulation was not in the national interest, and passed a number of laws between 1996 to 1998 to set a national framework for that industry, including the revised Telecommunications Act, the Internet Tax Freedom Act, and the Digital Millenium Copyright Act. Without that action, the Internet as we now know it would not have been possible.

The world will not pause for us. Congress should meet the moment by setting one national standard so American innovation is not governed by fifty different rulebooks.

Perry Metzger, Chairman of the Board, Alliance for the Future