Behind the Curtain: It's not too late

The White House, Congress and leading AI firms have allowed artificial intelligence to advance rapidly, without any serious attempt to regulate it or share its benefits beyond the wealthy elite. It's not too late to alter this course, but if they procrastinate, the risks of AI's worst-case scenarios could surge, possibly early next year, according to many private experts familiar with AI.

The significance lies in AI's rapid improvement, potentially leading to unpredictable self-improvement. On the positive side, there's a growing public demand for action and, in private, an understanding of feasible solutions with bipartisan and industry support.

In July, I wrote an article in The Wall Street Journal suggesting America faces a historic opportunity to collaborate on a generational challenge – winning the AI race. However, we're currently failing. Two months later, AI has advanced significantly, while our societal response remains inadequate. We're focusing on trivial matters instead of the looming threats posed by AI insiders.

To address this, I propose an all-hands-on-deck approach: sound an alarm for every political issue, even those seemingly minor. We must look back in shame if we trivialize AI concerns while ignoring the warnings of AI experts. While some may dismiss AI concerns as paranoia, the reality is that AI investment is unprecedented in human history, rapidly permeating every industry, job, community, and screen.

Instead of merely forming special sessions of Congress and select committees, we need more: a dedicated AI governing body with authority and speed to plan for scenarios and act promptly. Most major labs agree to this idea, provided it applies uniformly across competitors. This body would require the expertise to understand AI models and the authority to review and shut down any dangerous behavior. Moving at the speed of technology is crucial, not the typical regulatory pace.

A universal standard for AI data centers that benefit communities and taxpayers is another essential step. The top AI companies would readily accept this deal, as it allows them to build and operate data centers while providing community energy, offsetting water use, reporting transparently, and spreading financial benefits locally.

These companies are concerned about public backlash but are willing to share financial benefits through dividends or cash. We should leverage this opportunity to engage China in creating a worldwide AI framework similar to how nuclear weapons have been governed for decades. We need to start this discussion now to potentially collaborate in the coming years.

Lastly, building coordinating infrastructure is vital. A working group with actual authority, composed of experts from the federal government, leading AI companies, business and labor, economics, public health, and ethics, should be formed. This group would map potential problems and benefits, develop policies ahead of time, communicate with the American people, and expedite the delivery of widespread AI benefits.

While some, like President Trump, oppose regulation or bipartisan efforts, the political landscape is shifting, with Vice President Vance expressing growing concern. If this pace continues, we may soon face a pivotal moment in AI policy that was unimaginable just two months ago.