President Trump is pushing back hard against a growing chorus of AI safety advocates — including Anthropic CEO Dario Amodei — calling for a deliberate slowdown in artificial intelligence development, declaring that the United States is currently leading China in AI and that he intends to keep it that way, as Chinese President Xi Jinping prepares to visit the White House on September 24 for talks that will place AI prominently on the agenda.
The debate erupted this past week following a 3,800-word essay by Amodei warning that unchecked AI development carries mounting risks and proposing that the frontier be deliberately paced to allow safety to keep up with capability. House Speaker Mike Johnson echoed concern about AI's trajectory, acknowledging the need for "some guardrails, some safety measures" to ensure the technology "doesn't run away." But Johnson drew the same line as Trump on the China question, warning that any regulatory move that causes the U.S. to "lose our edge in the global competition with China" would constitute a national security threat. National Economic Council director Kevin Hassett confirmed the administration is convening meetings this week among cyber, science, and technology officials to discuss next steps — and that Trump may meet with concerned members of Congress to hear their concerns. The administration's posture is that AI leadership over China is the non-negotiable constraint around which any safety framework must be built.
"I want to keep it that way because whoever wins AI, wins. Whoever wins that race is probably going to win." — President Donald Trump, on the U.S. lead over China in AI development
The strategic context makes Trump's resistance to a slowdown straightforward to understand. Chinese AI labs — including DeepSeek, Zhipu AI, and Moonshot AI — have released a series of competitive low-cost, high-performance models in 2026 that match or approach OpenAI and Anthropic products on key benchmarks. U.S. technology export controls limiting Chinese companies' access to advanced AI processors — the chips needed to train the most capable models — remain the primary tool constraining China's development pace. China is simultaneously proposing a more open AI governance model internationally, positioning itself as the alternative to what Beijing characterizes as American AI imperialism. The Trump-Xi September 24 summit is expected to include discussion of AI guardrails, but analysts who track Chinese AI development have low expectations for any breakthrough given the two countries' broader geopolitical rivalry. AI has also become a significant domestic political issue with the November midterms approaching, as data center expansion into rural communities has generated bipartisan constituent concern over land use, water consumption, and electricity costs.
Former President Obama — now publicly engaged on AI as a campaign issue — urged Democrats to develop a concrete AI platform, saying he would address "what it means if there's going to be job displacement" in economically specific terms. House Minority Leader Hakeem Jeffries separately called for Congress to take "decisive action now" to slow development and protect the public — a position that puts progressive Democrats in the unusual position of arguing for restraint on a technology where the administration's critics more typically favor regulation.
The U.S. currently controls approximately 75% of global AI compute capacity; China controls roughly 15%. The gap reflects decades of private sector investment, semiconductor design leadership, and a regulatory environment that has allowed large-scale data center construction to proceed at pace. U.S. export controls restricting China's access to NVIDIA's most advanced chips — the H100 and subsequent models — have been the primary tool slowing Chinese AI development, though Chinese firms have made significant progress in developing domestic alternatives and in optimizing performance from less advanced hardware.
Trump's position is strategically sound and his instinct on the China competition is correct: the nation that leads in AI will have decisive advantages in economic output, military capability, and geopolitical influence for the foreseeable future. The harder question his administration has not yet answered is whether the speed of current AI development is itself creating risks that would undermine American leadership if they materialize — and whether a framework that builds safety into the race rather than slowing it down is possible. The Amodei essay and the Congressional concern it generated deserve engagement, not just dismissal. Winning the race matters. So does understanding what you've built when you cross the finish line.