The big topic on CNN this morning is an essay by Dario Amodei of Anthropic. On Saturday (Sept 12), Dario published an essay called “We Must Pace the Frontier,” where he argued the industry must slow the pace at which it improves AI model capabilities. This doesn’t mean stopping development altogether, but regulating it with high priority. Meanwhile, Donald Trump’s approach has been consistently deregulatory and “pro-industry”, favoring a light-touch, federally uniform framework over state-level rules.
In January 2025, Trump signed an executive order removing Biden-era AI regulations, aiming to “remove barriers to United States AI leadership.” His campaign platform had promised to repeal Biden’s AI order, calling it an imposition of “radical leftwing ideas” on AI development, in favor of an approach rooted in free speech. In July 2025 he issued his “America’s AI Action Plan” alongside three executive orders, with roughly 90 policy actions focused on accelerating innovation, building AI infrastructure, and increasing international AI diplomacy, thereby reconsidering existing regulations and suspending investigations seen as stifling innovation. Again in December 2025 he issued another executive order, preempting states in their efforts to hold back AI development…especially the building of data centers. The order aimed at significantly restricting states from independently regulating AI in “onerous and excessive” ways or in ways that conflict with federal priorities around AI innovation, leadership, and global dominance, with the stated goal of reducing “cumbersome” state regulation that “stymies innovation.” It also used federal funding as leverage, letting agencies condition discretionary grants on states not enforcing AI laws deemed inconsistent with the order. Then in March 2026 came his National AI Policy Framework. This called on Congress to preempt state AI laws seen as imposing “undue burdens,” while preserving state authority in areas like child protection, fraud prevention, zoning over AI infrastructure, and states’ own procurement rules. The framework aimed to create uniform national safety and security guardrails while blocking states from setting their own AI rules. What that really seemed to be was a political move to appease Republican outcry about data centers while keeping the AI development beast able to continue its juggernaut.
The bottom line is that Trump favors a single federal standard, minimal restriction, and blocking states from creating a “patchwork” of AI rules. This is essentially the opposite instinct from Amodei’s push in his “We Must Pace the Frontier” essay, which calls for third-party safety audits and coordinated slowdowns. Trump directly weighed in, and the tension between him and Amodei runs deeper than just this essay. On Thursday, September 10 (two days before Amodei’s essay published…he likely got advance notice), Trump said he does not “have any” concerns that the pace of AI development is too fast. To be clear, that is essentially the opposite instinct from Amodei’s “we must slow down” argument (agreed with now by both Sam Altman and Elon Musk). In discussing pacing more broadly, he’s emphasized maintaining America’s AI lead, and Amodei himself has argued that slowing down beyond where the U.S. currently leads could hand China an advantage. That means that even Amodei is framing his own proposal carefully around the China competition point Trump cares about most. Amodei has proposed international coordination.
Amodei’s proposal calls for several key elements. He wants labs to take enough time to align and safeguard models properly, verified by outside evaluators letting third-party evaluators confirm that safety work has actually been done. He has urged rival companies to invite embedded external auditors into their offices and coordinate a slowdown in development so regulators can catch up. And he’s asking for an antitrust waiver so competitors can legally coordinate on safety measures together. Sam Altman and Elon Musk, despite frequently being at odds with Amodei, publicly agreed, with Musk writing on X, “Dario is right”
AI has shaped up as the battle royale against humanity. The core technical worry for most of us is an AI system optimizing for a goal that’s subtly different from what humans actually want. That could allow AI to pursue those goals in ways that harm us, not out of malice but out of indifference to anything outside its objective. This is called the “paperclip maximizer” logic of competence without shared values. Fundamentally, this is a loss of control as systems get better at tasks like coding their own successors (what Amodei’s essay calls “recursive self-improvement”) or operating autonomously as agents (the concern being that human oversight becomes harder to maintain at each step, not that AI “wants” to fight humanity). A different worry, less about the AI itself and more about who controls it is that if a handful of labs or a handful of governments get decisively ahead in capability, that’s a human power struggle wearing an AI mask. Amodei himself has voiced discomfort with a few companies (including his own) making consequential decisions with no democratic mandate. The most immediate, and least sci-fi version of this concern is about AI competing with humans for jobs, resources (think power and water) and economic relevance. That doesn’t require any hostile intent, just capability plus market incentives.
It’s worth noting that this framing has critics that argue that the “AI vs. humanity” narrative is overblown or convenient. To me that sounds eerily like what people say about the breakdown of democracy under Trump. I would note that the other big item in the news is a piece in The New Republic by conservative judge J. Michael Luttig. He has published a dire warning to all Americans about a potential and impending breakdown in democracy. Luttig has outlined how Republicans could refuse to seat elected Democrats in January after a Democratic victory in the midterm elections. He calls it “The Final Battle for America’s Democracy.”
Let’s face it, we have had ten years of persistent erosion to democracy. Democracy has become the boiling frog. Trump has kept raising the temperature of the water at the sake of some commercial transactional reason or another. Luttig is just telling us that frog is about to be on the menu. It seems to me that Trump is not satisfied just putting money ahead of democracy, he now wants to put money ahead of humanity altogether. He may finally have met his match. Where he’s gotten the best of democracy, I’m betting that humanity will kick his AI ass in the ultimate cage match!