Core Summary
The Guardian reported on September 13 that OpenAI CEO Sam Altman and Elon Musk, along with other AI industry leaders, made a rare joint call to slow “reckless” artificial intelligence development, warning that frontier models may pose systemic risks to human civilization. Almost simultaneously, senior officials in the Trump administration publicly rejected any slowdown, arguing that excessive regulation would undermine U.S. global competitiveness in a critical technology. The open disagreement between industry leaders and the White House marks a new phase in the long-running U.S. debate over whether to accelerate or decelerate AI development, and is likely to shape legislative pace, industry self-regulation and cross-border regulatory coordination for the foreseeable future.
Event Details
A rare joint statement from industry leaders. On September 13, OpenAI CEO Sam Altman, Elon Musk and Anthropic CEO Dario Amodei, among other industry leaders who have long competed and litigated against each other, made a rare joint public appeal, in statements and on social media, to introduce “necessary deceleration” into frontier AI development. Their statement argued that the training scale, capability leaps and deployment scope of some frontier models have already exceeded the carrying capacity of existing safety assessments, industry self-regulation and regulatory frameworks, and called for internationally agreed compute thresholds, transparency disclosures and red-line mechanisms.
The Trump administration’s clear rejection. Senior White House officials said at a press briefing later the same day that they would not slow the pace of AI development. They argued that artificial intelligence is one of the most strategically valuable technologies of the century and that over-regulation or unilateral deceleration would cost the United States its lead against China, the European Union and other major economies. President Trump also said in an interview that he would not change his view that AI risks are “overblown,” and that the United States would continue to drive technology adoption and industrial deployment through market mechanisms.
The substance of the disagreement. Industry leaders argue that frontier models already have the potential to exert irreversible influence across cybersecurity, the biological sciences, financial markets and the information ecosystem, and that mandatory assessments and external review must be installed at the key inflection points of capability leaps. The Trump administration holds that current AI risks can still be addressed through existing laws, industry standards and market mechanisms, and that a mandatory slowdown mechanism would drag on innovation and indirectly weaken U.S. national security and industrial advantages.
Anthropic researchers had already spoken out. Just before this joint statement, a senior researcher at Anthropic publicly resigned, saying she could no longer work inside what she described as an “out-of-control” AI capability race. Her remarks, which spread widely across Silicon Valley and the global tech media, have been seen as an important inflection point in industry self-reflection on the pace of frontier development.
Reactions across the legislative and regulatory landscape. Multiple AI-related bills in the U.S. Senate and House have moved in different directions since the statement. The Senate majority leader said the chamber would push forward in the coming months on transparency disclosures, compute registries and mandatory assessments for high-risk models. The relevant House committee, by contrast, prefers an approach centered on industry self-regulation, voluntary commitments and inter-agency coordination, and opposes any mandatory slowdown provisions.
International reverberations. Regulatory bodies in the European Union, the United Kingdom, Japan and South Korea have all weighed in on the pace of AI governance over the past week, with most supporting a baseline international standard for risk assessment. Chinese officials, in a recent policy briefing, also stressed the need to advance AI safety assessments and cross-border coordination, while reiterating China’s commitment to an autonomous and controllable development path.
Full Perspective
The open disagreement between OpenAI, Musk and the Trump administration over whether to slow the pace of artificial intelligence development is not a simple policy debate. It is a multi-layered contest over the trajectory of technological evolution, the structure of industrial competition, the logic of national security and the architecture of global governance, and its outcome will reshape the global technology landscape for years to come.
The AI industry is moving from competition to conditional cooperation. Over the past several years, OpenAI, Musk-linked AI entities and Anthropic have been locked in commercial competition, lawsuits and a bidding war for talent. Their joint call to slow “reckless” development is the first time these firms have taken a unified position on a major governance issue, suggesting that frontier AI development itself has reached a red line of shared concern among key industry players. The coordinated statement is likely to become an important opportunity for building cross-company safety standards and joint assessment mechanisms.
The Trump administration’s “no slowdown” stance will further fragment global AI regulation. The United States has chosen a market-driven, industry-first path, while the European Union, the United Kingdom, Canada and Japan are inclined toward stricter mandatory assessments and transparency disclosures, and China is pushing safety assessments and international coordination within an autonomous and controllable framework. This divergence of paths will force multinational AI companies to comply with multiple regulatory regimes simultaneously, significantly raising operating costs, compliance complexity and cross-border data-flow risk.
Frontier model capability leaps are approaching a social tolerance threshold. Multiple research institutions and industry leaders have repeatedly warned in recent months that frontier large models are already approaching the tolerance threshold of existing law, social norms and industry conventions in cybersecurity offense and defense, automated biological experimentation, financial market manipulation and the reshaping of the information ecosystem. Without mandatory assessments and external review, irreversible negative events could occur simultaneously across multiple domains within the next few years, triggering a systemic collapse of public trust in AI.
The tension between national security logic and industrial competition logic is deepening. From a national security perspective, frontier AI has become a core arena of great power competition, and excessive public evaluation may expose capability boundaries to rivals. From an industrial competition perspective, however, the absence of unified safety standards amplifies systemic risk and pushes companies into a prisoner’s dilemma in which whoever slows down first falls behind. How to balance national security, industrial competition and technological ethics is the central challenge of U.S. AI governance in the coming years.
The window for cross-border governance coordination is narrowing. The EU AI Act, the U.K. AI Safety Summit, U.S. executive orders and China’s related administrative measures together form the current “four-pillar framework” of global AI governance. The split between industry leaders and the White House highlights the fact that the major players still lack consensus on risk assessment methodology, compute thresholds, transparency disclosure and red-line mechanisms. If a minimum degree of international coordination cannot be established within the next 12 to 18 months, global AI governance may enter a period of heightened uncertainty.
Multiple Viewpoints
Industry leaders at OpenAI, Anthropic and in the Musk camp believe that frontier AI already has the potential to exert irreversible influence across multiple domains and are calling for internationally agreed compute thresholds, mandatory assessments and transparency disclosure mechanisms to systematically slow the “reckless” pace of development, arguing that this is a necessary choice for the entire industry to earn long-term social trust.
The Trump administration and senior White House officials insist that AI risks are overblown and that mandatory slowdown provisions would undermine U.S. global competitiveness in critical technologies, potentially ceding the lead to strategic rivals. They favor a market-driven, industry-self-regulating and inter-agency coordinated approach to governance, and oppose mandatory slowdown clauses at the legislative level.
The U.S. Senate majority leader leans toward supporting AI transparency disclosures, a compute registry and mandatory assessments for high-risk models, arguing that the legislative pace must keep up with the speed of frontier model capability leaps. Senators have noted that the absence of unified federal standards could leave states, agencies and companies caught in a fragmented compliance landscape.
The European Commission, the U.K. AI Safety Institute and Japan’s Ministry of Internal Affairs and Communications have recently expressed support for an internationally agreed baseline standard for risk assessment, stressing that cross-border coordination is the key to avoiding regulatory arbitrage and industry offshoring. Multiple bodies have argued that the risk of “fragmentation” in AI governance may be greater than the risk of “overregulation.”
Anthropic researchers and academia broadly call for greater transparency and interpretability in frontier model development, requiring companies to subject key aspects such as training data, compute scale, reward mechanisms and capability evaluations to independent third-party review. Prominent scholars argue that voluntary industry commitments are no longer sufficient to keep pace with current capability leaps, and that legislative and cross-border coordination mechanisms are essential.
Chinese institutions such as the Chinese Academy of Social Sciences, Tsinghua University and the China Academy of Information and Communications Technology have stressed in recent briefings and reports that China will adhere to an autonomous and controllable development path while promoting AI safety assessments, cross-border coordination and ethical governance. Chinese scholars argue that global AI governance should not be captured by a single country’s industrial policy, and that a more resilient coordination framework should be built on a multilateral basis.
编辑：GoodInfo全球资讯组