Trump's AI Risk Dismissal Puts China Race Above Safety Rules

President Donald Trump has moved to dismantle key artificial intelligence safety frameworks established by his predecessor, framing regulatory caution as a strategic liability in the accelerating technology rivalry with China. The policy shift, confirmed by White House officials and reported by Reuters and the Associated Press, signals that Washington intends to prioritise speed and commercial dominance over risk mitigation — a posture that analysts warn could have profound consequences for democratic governance, civil liberties, and international AI norms.

The Strategic Logic: AI as a Geopolitical Weapon

At the heart of the Trump administration's recalibration is a doctrine that equates safety regulation with competitive surrender. Senior officials have argued, both publicly and in internal communications obtained by journalists, that requiring AI companies to submit models for federal safety review creates delays that China's state-backed AI sector does not face. In Beijing, companies such as Baidu, Alibaba, and the military-linked research institutions operating under the PLA's Science and Technology Commission face no equivalent independent oversight.

The Race Narrative and Its Limitations

The framing of AI development as a zero-sum race has gained significant traction in Washington's policy circles, particularly following the emergence of DeepSeek, a Chinese AI model that demonstrated performance rivalling leading American systems at a fraction of the reported cost. That disclosure sent shockwaves through Silicon Valley and prompted an urgent reassessment on Capitol Hill. According to Foreign Policy, the DeepSeek episode was cited directly by administration officials as justification for accelerating deregulation.

However, AI researchers and ethicists have challenged the logic. The argument that removing safety guardrails produces faster, superior AI is contested by substantial technical evidence. Systems deployed without adequate red-teaming and alignment testing are more likely to produce dangerous outputs, hallucinate at scale, or be weaponised for disinformation — outcomes that, analysts note, ultimately undermine rather than strengthen national competitiveness. For further context on how Washington's broader regulatory posture is affecting the technology sector, see our reporting on Trump's AI controls shift putting Silicon Valley on edge.

What Safety Frameworks Were Removed — and Why It Matters

The revoked Biden-era order contained several distinct pillars. It required companies training AI models above a defined computational threshold to notify the federal government and share safety evaluations. It mandated that the National Institute of Standards and Technology (NIST) develop new standards for AI risk. And it directed federal agencies to assess AI's impact on critical infrastructure, including the power grid, financial systems, and healthcare networks.

The NIST Rollback and Its Consequences

The administration's budget proposals, as reported by AP, significantly reduce funding for NIST's AI Safety Institute — the body charged with developing technical benchmarks for model evaluation. Without those benchmarks, both government procurement officers and private sector actors lose the standardised vocabulary for assessing whether an AI system is safe to deploy in sensitive environments. The UN Secretary-General's office, in a recent advisory report on digital governance, warned that the erosion of national AI safety institutions in major democracies creates a vacuum that is unlikely to be filled by international bodies in the near term (Source: United Nations Advisory Body on AI Governance).

The legal landscape has grown more complex in parallel. A recent judicial ruling found that the Pentagon retaliated against AI safety-focused firm Anthropic, raising urgent questions about how the government treats companies that publicly advocate for caution. Our analysis of that case is available in our coverage of the judge's ruling that the Trump Pentagon retaliated against Anthropic.

China's AI Posture: A Reality Check

The administration's narrative depends heavily on the premise that China is racing ahead unconstrained. The reality is more nuanced. Beijing has, in fact, introduced its own AI regulatory framework — including rules on generative AI content moderation, algorithm transparency requirements for recommendation systems, and mandatory registration for large language model providers. These regulations are not safety-oriented in the Western civil liberties sense; they are primarily designed to ensure political compliance and state control of information flows.

Military Applications and the Taiwan Variable

Where China's AI development intersects most acutely with US national security concerns is in military applications. The PLA has been actively integrating AI into command-and-control systems, autonomous drone swarms, and electronic warfare platforms. That militarisation drive is inseparable from Beijing's intensifying pressure on Taiwan. Readers following that dimension of the rivalry can find broader analysis in our reporting on how China is increasing military pressure on the Taiwan Strait.

Security analysts caution, however, that the United States maintaining AI safety standards for civilian applications does not inherently weaken its military AI posture. The Pentagon's own AI programmes operate under separate authorities and are not subject to the civilian safety frameworks that the administration has dismantled. The conflation of civilian AI safety rules with military AI competitiveness, critics argue, is a category error that serves a deregulatory agenda rather than a coherent strategic doctrine (Source: Foreign Policy).

Implications for the United Kingdom and Europe

The US policy shift carries direct and significant implications for the United Kingdom and the European Union. Both have invested considerable political capital in positioning themselves as credible, safety-conscious AI governance leaders — a posture that now faces pressure from two directions simultaneously.

The UK's Delicate Balancing Act

London's approach to AI regulation has been deliberately light-touch — relying on existing sectoral regulators rather than creating binding primary legislation — in part to signal to American technology firms that Britain is a friendlier destination than Brussels. The UK's AI Safety Institute, launched with considerable fanfare, sought to build a transatlantic partnership with its US counterpart under the Biden administration. That partnership's operational continuity is now in question following Washington's deregulatory pivot, according to Reuters.

For British policymakers, the dilemma is acute. Maintaining rigorous AI safety standards risks being framed as falling behind in the US-China race — a characterisation the Trump administration will not hesitate to make in trade negotiations. Abandoning those standards risks a race to the bottom that the EU AI Act, which British firms operating in European markets must still comply with regardless of Brexit, explicitly prohibits for high-risk applications.

The broader pattern of the Trump administration subordinating institutional integrity to political priorities is visible across multiple domains. The recent controversy over federal museum funding, detailed in our reporting on Trump's Smithsonian order putting federal museum funding at risk, illustrates the administration's willingness to redirect federal resources away from bodies that operate with any degree of independence.

European Regulatory Sovereignty Under Strain

The EU AI Act, which entered into force this year, represents the world's most comprehensive binding AI regulatory framework. It classifies AI systems by risk tier — from minimal to unacceptable — and imposes conformity assessments, transparency requirements, and human oversight obligations on high-risk deployments in areas including healthcare, education, employment, and critical infrastructure. American technology companies selling into European markets remain subject to these obligations regardless of what Washington does domestically.

European Commission officials have indicated they have no intention of weakening the AI Act in response to US pressure, according to reporting from AP. However, there is growing anxiety in Brussels that a significant divergence in US and EU regulatory standards could create a two-tier global AI market — with American companies developing looser, faster products for domestic and allied markets while maintaining separate compliance tracks for European deployment (Source: AP).

The Broader Democratic Stakes

Beyond the immediate US-China framing, analysts and civil society organisations argue that the dismantling of AI safety frameworks poses risks that transcend geopolitical competition. AI systems deployed without adequate testing have already produced documented harms in hiring algorithms, bail-risk assessments, and facial recognition policing tools. Scaling these systems further, faster, and with less government oversight does not merely risk economic inefficiency — it risks embedding discriminatory, opaque, and unaccountable decision-making into critical social institutions.

The UN Advisory Body on AI Governance, in its most recent report, identified the absence of pre-deployment safety evaluations as a primary systemic risk in democratic societies — noting that the harm from inadequately tested AI systems disproportionately falls on marginalised communities with least political capacity to demand accountability (Source: United Nations Advisory Body on AI Governance).

Foreign Policy has noted that the Trump administration's AI posture is consistent with a broader philosophical disposition: that governmental oversight is inherently a constraint on American dynamism rather than a guarantor of public trust. That argument, however persuasive it may be to deregulatory ideologues, does not address the fundamental asymmetry — the United States is not merely competing with China on AI capability; it is also competing on the global stage for the argument that democratic governance produces better, more trustworthy technology. Gutting the institutions designed to ensure that trustworthiness weakens that argument precisely when it matters most.

As the global AI governance landscape splinters between Washington's deregulatory turn, Brussels' binding rulebook, and Beijing's state-control model, the coming months will test whether the United Kingdom — positioned awkwardly between all three — can construct a coherent national AI strategy that preserves safety without sacrificing competitiveness. The window for that effort, officials and analysts alike acknowledge, is narrowing rapidly.