In the News

America needs a strong national strategy on artificial intelligence, and we need it now. AI is moving at breakneck speed, producing newer and more powerful models that pose real risks to safety, national security, and the nation’s workforce. Each day that passes without a comprehensive federal standard puts us further behind, and puts every American at greater risk.

The promise is enormous, nowhere clearer than here in Massachusetts. Our universities, hospitals, and startups are using AI to hunt for cancer cures, detect disease earlier than any doctor could, and model the climate threats bearing down on our coastlines. The Commonwealth’s researchers, life-science labs, and cybersecurity firms should lead the world in building this technology.

However, the same technology that could cure disease can also cause profound harm. Parents are seeing chatbots cause real harm to their children, workers are worried about their job prospects, and educators are struggling to find healthy ways to implement this new technology in class.

In the absence of a federal regulatory framework, power concentrates in the hands of a small number of companies racing to build the most potent technology in human history. They ask us to trust them while they write the rules of the road themselves. We’ve seen this movie before. Lawmakers spent the last two decades playing catch-up to social media giants, and the public is still paying the price.

Recognizing the lack of federal action, states have stepped in to fill the void. States like California, New York, and Illinois have passed laws to target potential catastrophic risks when AI models are developed. Massachusetts lawmakers are targeting the harms of AI usage, including by chatbots and AI-generated sex abuse imagery. This is important progress, and the leadership of our state lawmakers has finally spurred the early stages of a federal response.

Last week, the White House issued an executive order creating a voluntary system for evaluating powerful new AI models. The administration’s engagement is a positive step, but a voluntary pledge holds only as long as these massive companies choose to play along. Only Congress can set a durable nationwide standard with the force of law behind it.

That’s why I, a Democrat, entered into bipartisan negotiations with Republican Representative Jay Obernolte of California to craft federal legislation that governs the development of AI models. We have four clear goals: ensure AI models are trained safely, guarantee that states can continue passing and enforcing laws to defend their residents from harms when models are deployed or used, arm policy makers with the information necessary to respond in real time to AI’s impacts across the workforce, and focus the regulations on the largest AI companies while giving startups and small businesses the space to build.

Together, we crafted the Great American Artificial Intelligence Act, a discussion draft that creates two distinct lanes. It tasks the federal government with setting uniform safety rules for how the most powerful models are developed and tested, a federal standard that, for three years, would take the place of a handful of state laws governing the development stage so innovators face one rulebook, not 50. States, meanwhile, would retain full power to govern how AI is deployed and used, from hiring and housing to health care, education, chatbots, and more.

This governance model is tried and tested. When cars are built for sale in the United States, they must meet federal safety standards, so a car sold in one state is as safe as one sold in another. Once those cars hit the road, however, states write the rules of the road. No one wants Washington setting the height limit for moving trucks on Storrow Drive, but the federal government should make sure every vehicle that drives on it is built to keep drivers and families safe.

A patchwork of 50 different state laws cannot protect our national security, and it does not foster the innovation we need to compete. It leaves startups in Cambridge and Devens navigating a maze of conflicting rules while our adversaries race ahead. A strong federal standard protects our country, gives innovators the certainty they need to grow, and ensures state legislators can pass laws so AI products meet the needs of their residents.

This is a difficult balance. Some argue the bill is too burdensome, others that it doesn’t go far enough. That’s why we released it as a discussion draft to gather input from workers, researchers, educators, startups, state and local lawmakers, industry, and the public, all of whom have a stake in getting this right.

We’re at a pivotal moment. AI models will only grow more capable, and the glaring risks will only grow with them. The Great American Artificial Intelligence Act is meant to help unlock a comprehensive, bipartisan federal AI strategy, one that lets Massachusetts lead.