Anthropic CEO Dario Amodei ignited a fierce debate this weekend with his essay “We Must Pace the Frontier,” arguing for deliberate slowing of AI development to ensure safety. He proposed three concrete steps: embedding third-party evaluators to verify safety practices, coordinating standards among democratic frontier AI companies, and attempting global pacing agreements with authoritarian regimes where possible. Amodei clarified that pacing isn’t halting progress but ensuring adequate time for alignment. Crucially, he announced Anthropic is unilaterally committing to the first step, inviting external scrutiny into its operations.
OpenAI’s Sam Altman quickly endorsed the core premise, stating that pacing has been a primary internal topic recently. He agreed with the need for independent evaluators and welcomed a federal framework for consistent safety requirements. Altman emphasized that shared standards for misalignment and monitoring would yield better outcomes, warning against two extreme failures: losing control to AI or creating excessive power concentration. Google DeepMind’s Demis Hassabis also backed the direction, noting it aligns with frameworks he previously published. Even Elon Musk reposted the essay with a simple affirmation: “Dario is right.” This rare consensus among rival CEOs suggests industry leaders see regulatory guardrails as inevitable, if not desirable.
The political response, however, fractured sharply along partisan lines. President Donald Trump rejected the slowdown narrative on Truth Social, claiming the only necessary guardrail is a “STRONG AND SMART” president. He accused Amodei of pretending to be a “perfect little angel” while suggesting a conspiracy against AI benefits China. Vice President JD Vance echoed this skepticism, calling industry requests for regulation a “trojan horse” amidst the race with Beijing. House Speaker Mike Johnson struck a middle ground, urging meetings to balance national security with model safety but ruling out any moratorium that might let China overtake the US. Nvidia’s Jensen Huang, speaking alongside Trump, reassured audiences that robots won’t take over, focusing instead on broad economic wins.
This divergence reveals a fundamental tension: tech leaders view pacing as a risk mitigation strategy, while political figures frame it through a lens of geopolitical competition. The industry’s willingness to invite oversight contrasts with political rhetoric that equates regulation with weakness. As Amodei’s proposal gains traction among peers, its viability depends less on technical feasibility and more on whether Washington views safety pauses as prudent caution or strategic surrender. The coming months will test if these unified corporate voices can withstand the pressure of an accelerating arms race.
