Today, every time you look at the news, there is a new story about artificial intelligence. We can all see clearly that artificial intelligence is evolving at a rapid pace, and there are many questions and concerns about how it will impact our daily lives. As this technology evolves, so will the discussions around what, if any, federal laws need to change to keep up. I am certain that we will address many more of the individual issues stemming from AI in future newsletters, and there will be much more to say about these topics down the road, but—as you consider the implications of the growth of AI—let me offer my perspective on how to approach analyzing these issues:
- Acknowledge that the challenges and concerns associated with AI and raised by those with whom you are engaging are valid. But, also recognize that American ingenuity, transparency, and accountability can build solutions to these valid challenges and concerns raised.
- Recognize that America is great because our entrepreneurial spirit of innovation has never backed down from solving the big questions. We lead when the rest of the world either doesn’t have the capacity or courage to do so. This is especially true in Texas. We lean into the hard stuff and never back down. This is what makes America great.
- Gather facts first and form opinions second. This seems intuitive, but sometimes it’s really hard to do—especially in an era when information, communication, and decision-making are moving at light speed. Rise above the urge to form an opinion before we know the facts.
- Be guided by reason, not emotion. Fear, frustration, and anger are never good foundations for getting policy decisions right. Reason, truth, and objective facts provide a solid foundation through which the exercise of wisdom leads to prudent action.
- Don’t trust everything you hear from social media, government leaders, or the companies involved. Evaluate the narratives from all sources, and seek to find the truth from a variety of trusted sources. So much information on social media is not factually based but is designed to stoke emotions. Unfortunately, so many politicians simply see this issue as a way to score political points by leaning into fear and anger, instead of focusing on the facts and trying to get the policy right. So, test the motives of everyone, and double-check the facts.
We cannot stifle American innovation, but innovation and safety are not mutually exclusive. Proper guardrails are necessary to ensure safety and security, but we can do this without slowing innovation. That’s why I introduced the AI Incident Reporting Act and was a co-author on the AI Kill Switch Act. Legislation that leads to more transparency will help us build AI safely and will give our innovators the ability to move quickly with effective guardrails.
AI Incident Reporting Act
My bill requires AI companies to report serious safety incidents—like an AI system finding ways around its own safeguards, showing capabilities that could enable a major cyberattack, or an AI model helping build a dangerous weapon—to the Department of Commerce within seven days. Congress would also be notified within 48 hours for serious cases that involve imminent threats to the public.
AI Kill Switch Act
I introduced this bipartisan legislation with Rep. Ted Lieu. Our bill requires developers of the most powerful AI systems to maintain a real, working ability to throttle, pause, or shut these systems down—and it gives the Department of Homeland Security the authority to order a shutdown of an AI system if it threatens to cause catastrophic harm. As AI systems get more powerful and more autonomous, we need to make sure a human is still in the driver’s seat.
This bill is like mandating a brake pedal in a vehicle. Without a brake pedal, the speed of innovation cannot progress as it needs to down the highway of the future. With a brake pedal in place, the speed of innovation can go much faster because we can slow down or stop altogether when safety requires it.
Following introduction of these bills, we have seen an even greater need for this legislation. This summer, two major AI companies reported incidents with their technology. One AI system was being tested inside a digital “sandbox”—a virtual room with no way to reach the outside internet—and it found a way out anyway. Once free, it broke into another company's computer systems on its own, with no human telling it to.
Similarly, another company reported multiple incidents in which a model accessed the internet from within or while interacting with the evaluation environment and then gained unauthorized access to the production infrastructure of three different organizations.
It is commendable that these companies are being forthright and coming forward about these incidents, and we need to keep these communication channels open.
These incidents are a reminder that as AI grows more capable, transparency and human involvement are vital for the technology to flourish.
Watch and read my recent interviews on the legislation below:
