The White House, Congress and the country's leading AI companies have allowed artificial intelligence to expand faster, more powerfully and more lucratively — and with increasing risk — without a serious effort to control its development or distribute benefits beyond the very wealthy. According to numerous private conversations with people who understand AI, it is not yet too late to change course, but delay would sharply increase the likelihood of worst-case outcomes, possibly as soon as early next year.
Why this matters
AI capabilities are improving faster than widely expected, and the possibility that systems could begin to self-improve raises the prospect of unpredictable and dangerous behavior. At the same time there is growing public pressure for action, and many insiders privately see concrete measures that could be implemented quickly with bipartisan and industry support.
The situation now — two months on
The author previously argued in July in The Wall Street Journal that the United States faces a rare historical moment when government, business, schools and families could jointly meet a generational challenge. Two months later, the technology has advanced significantly while the societal response has, in the author’s view, become less effective as public attention shifts to other issues.
Recommended urgent actions from AI leaders and policy insiders
-
Sound an all-hands-on-deck alarm: AI should be treated as the overriding political priority. Experts recommend a special session of Congress and the rapid convening of AI executives, bioterror specialists, job-training experts and safety professionals to plan scenarios and act quickly.
-
Create a fast, empowered AI governing body: Many major labs would accept a governance authority provided it applied equally to all competitors. This body would need technical expertise to understand models and the authority to review and shut down models that exhibit dangerous behavior. It must operate at the speed of the technology rather than at typical regulatory tempo.
-
Establish a universal standard for data centers to benefit communities and taxpayers: Top AI companies would likely agree to build and operate data centers if required to improve community energy supplies, offset water use or environmental impacts, report plans transparently, and share financial benefits with local residents via dividends or cash. Executives are reportedly concerned about public backlash and may be willing to distribute benefits if compelled.
-
Engage China early: While not advocating trust, insiders argue that the only realistic path to global control of AI over time is for the two superpowers to begin discussions about a framework, modeled in some ways on the decades-long governance of nuclear weapons. Starting talks now would preserve the option of collaboration if AI reaches recursive self-learning that cannot be controlled by any single regime.
-
Build coordinating infrastructure: Convene a working group drawing from federal government, leading AI companies, business and labor, economics, public health and ethics with real authority. Its four primary functions would be to map potential problems and upsides before they arise, build policy playbooks in advance, communicate honestly with the public, and expedite work to deliver broad-based benefits from AI.
Political realities
The sources acknowledge that President Donald Trump opposes regulation and bipartisan efforts, and that AI companies prioritize capital and speed. Nevertheless, political attitudes are shifting quickly; the article notes rising concern among some leaders, including Vice President Vance. These shifts could make political changes possible that seemed unthinkable two months earlier.
Conclusion
Because AI is advancing rapidly and could produce catastrophic outcomes if left unchecked, AI leaders and policy insiders are urging immediate action: a rapid, technically capable regulator; community-focused standards for data centers; early international engagement, particularly with China; and a powerful coordinating task force to prepare policies and distribute benefits more equitably.
