Trump and Mike Johnson reject calls to slow down AI

Silicon Valley's biggest rivals just managed something close to a miracle: agreeing on something. Washington, however, was not impressed.

Anthropic CEO Dario Amodei published an open letter on September 12 titled "We Must Pace the Frontier," urging the industry to intentionally slow the development of the most advanced AI systems. His plan proposed embedding outside evaluators inside AI labs, coordinating rules among democratic nations, and pursuing global agreements.

Remarkably, the proposal quickly won over his fiercest competitors. OpenAI CEO Sam Altman backed the call, committing to giving outside auditors employee-level access. Elon Musk simply posted, "Dario is right," while Alphabet’s Demis Hassabis offered tentative support.

Washington wants the pedal down

That executive solidarity hit a brick wall in Washington almost immediately.

Speaking to the Financial Times, Donald Trump dismissed the notion of slowing down, warning that "whoever wins AI, wins," and that the U.S. must preserve its lead over China. According to AP, Trump also called fears of AI ending humanity a "HOAX" on Truth Social, comparing them to climate warnings and insisting that a "High IQ" president is the only safeguard the country needs.

House Speaker Mike Johnson took a similar stance on CNN. Talking to Jake Tapper, Johnson warned that rushing to restrict frontier systems would be a "national security threat." He cautioned against panic, arguing that if Congress holds an emergency session to regulate AI, the U.S. will simply lose the race to Beijing.

A geopolitical deadlock

Pushback also came from abroad. Amodei's proposal argued that Chinese leadership in AI threatens U.S. security and advocated keeping advanced AI chips out of China's hands.

At a September 14 briefing, Chinese Foreign Ministry spokesperson Guo Jiakun pushed back against "fear-mongering, confrontation and vicious competition," advocating instead for open, inclusive governance. State-run outlet Global Times had already labeled the chip restrictions a "Cold War playbook" in an editorial a day earlier.

While Trump and Xi Jinping have a meeting planned for September 24 where AI governance is expected to come up, finding common ground looks extraordinarily difficult. Trump rejects catastrophic risk claims, while Beijing rejects technology containment. Amodei himself acknowledged that broad slowdowns may be unrealistic, pointing to biological weapon safeguards and pre-release testing as more practical starting points.

Why it matters

We think this standoff exposes the brutal game theory behind cutting-edge AI.

For months, labs have defended reckless release cycles by claiming they couldn't afford to fall behind. Now, the people actually building the frontier—Amodei, Altman, Musk, Hassabis—are publicly admitting that the models are advancing dangerously fast. The technical elite wants guardrails.

Politics, however, has effectively closed the door on a coordinated, government-backed pause. With Washington viewing every safety delay as a gift to China, no lab can slow down without fearing it is surrendering its market lead. Antitrust waivers and federal mandates to halt training runs are simply not happening anytime soon.

That puts the ball right back in Silicon Valley's court. If politicians won't mandate safety, the test now is whether these labs will uphold their voluntary pledges. Altman promised outside evaluators employee-level access, and Amodei asked for auditors who can publish critical findings. Whether they actually open their doors—and listen when red flags are raised—will show if this was genuine alarm or just high-minded theater.