Introduction
The AI policy and governance philosophy of the United States have diverged dramatically from those of the European Union in the first half of 2025. While the EU emphasizes comprehensive, rights- and risk-based AI legislation through the AI Act, the US AI Plan’s deregulatory approach prioritizes technological dominance over safety constraints, creating fundamentally incompatible governance models. Concurrently, extremely rapid advances in AI-based capabilities, particularly those powered by Generative AI (such as ChatGPT, Claude, Gemini, Llama), are driving a push to integrate AI throughout most aspects of society. Estimates on AI’s contribution to global GDP vary widely but easily reach $20 trillion USD [1]. However, the uncertainty and variance in international standards of regulation and governance poses a major barrier to AI adoption [2].
In recent years both the US and the EU have recognized that there are inherent risks in the use of AI, related to issues such as data use, accountability, fairness, and transparency. To address these risks, they put into place regulations governing AI development and use. However, 2025 changes in US policy have moved to deregulate in favor of rapid technological dominance. Additionally, the EU AI Act looks to provide comprehensive legislation and penalties, while the US approach, as expressed in the AI Action Plan [3] aims to “revise, or repeal regulations, rules, memoranda, administrative orders, guidance documents, policy statements, and interagency agreements that unnecessarily hinder AI development or deployment.” These two guiding philosophies now differ enough to cause greater uncertainty in the business environment. This impacts investment and AI development decisions, and increases the barriers to adoption for multinational corporations.
EU Approach: Comprehensive Legislative Framework
The European Union’s AI Act [4], effective on August 1, 2024, represents the world’s first implementation of comprehensive AI regulation. It categorizes AI systems into four risk levels: unacceptable, high, limited, and minimal risk, with different requirements for each category. Unacceptable risk AI applications like social scoring (e.g., AI using individuals’ social behaviors to determine their treatment in an unrelated context) are banned outright, while high-risk applications (i.e., where AI may affect citizens safety, and fundamental rights) face strict legal requirements. All high-risk applications must be assessed before market entry and throughout their lifecycle.
It has been just over one year since the AI Act entered force. The year has revealed compliance challenges for small/medium business, incomplete conceptual definitions, and challenges converting principles to practice. At the same time, advances in AI and unforeseen, novel applications threaten to exceed the Act’s initial regulatory scope [5].
US Approach: Competition over Regulation
The US fundamentally reversed its AI policy in 2025, from rights-protection to competition, prioritizing technological dominance over regulatory oversight. The January 2025 presidential executive order, “Removing Barriers to American Leadership in Artificial Intelligence,” explicitly rolled back AI policies that the administration felt would stifle private sector innovation and threaten American technological leadership.
The recent centerpiece policy, “America’s AI Action Plan” states that “the United States must win the AI race” to remain the leading economic and military power. The plan focuses on promoting American technology dominance through three pillars: accelerating innovation, building AI infrastructure, and leading in international diplomacy and security. This approach emphasizes the private sector’s role while reducing bureaucratic red tape, creating a clear deregulatory contrast to the EU’s comprehensive legislative framework. However, here also, vague or subjective terms (like “ideological neutrality”) create operational and legal uncertainty for vendors. Additionally complicating matters, however, is that state and local governments are already legislating on AI[6].
Timing and Enforcement
Beyond philosophical differences, there are concrete implementational differences of note. Foremost, enforcement of regulations is clearly spelled out in the EU AI Act, with severe penalties (up to €35 million or 7% of global annual turnover for the most serious violations) in place for non-compliance and the prohibition of certain AI practices. In contrast, the US AI Plan mentions enforcement only in the context of export controls, and federal funded biosecurity research.
Additionally, The EU AI Act has been designed for a phased rollout with mandatory deadlines over the timespan from August 2024 to February 2027, allowing time for business to adapt to the new environment. In contrast, the US AI plan directs immediate action to remove any barriers established by the previous executive orders.
Finally, both approaches to AI regulation are subject to risks. As noted earlier, AI technologies are advancing quickly, and the EU AI Act must contend with technologies and applications unforeseen during its drafting. The Act must be continually reexamined, to keep up with or lead the technology, rather than chasing it. On the other hand, the US approach, with competition as its primary value, risks accumulating significant technological debt, as the lack of proactive protection through regulation risks safety and ethical consequences not easily corrected after the fact.
Implications of Divergent Approaches
This regulatory divergence creates significant global governance challenges by fragmenting international AI standards. The US deregulatory approach may accelerate innovation at the cost of rights protection, while the EU’s precautionary model may protect consumer rights but could disadvantage European AI competitiveness. Other major world economies will choose between these two, or find their own way, creating competing regulatory blocs that complicate cross-border AI deployment and undermine efforts to establish unified global AI governance frameworks.
Multinational AI companies must navigate these conflicting compliance requirements, driving up development costs and complexity. Companies face difficult strategic choices: develop separate AI systems for each jurisdiction, potentially doubling development expenses, or build to the higher EU rights standards and sacrifice competitive advantages in the deregulated US market. This fragmentation may also induce regulatory arbitrage, where companies may relocate development to favorable jurisdictions. It will also complicate partnerships, mergers, and global AI product launches, as companies must assess compatibility across fundamentally different environments.
Navigating the Mixed Regulatory Environment
Under the current regulatory divergence, multinational AI companies must adopt dual-track strategies, developing EU-compliant systems with robust documentation, transparency, and risk assessment protocols while maintaining streamlined versions for the US market. Best practices should include respecting and implementing privacy and responsible AI standards as a baseline to satisfy EU requirements as well as deliver safer systems in all markets. The consequences of failure in high-risk use cases may not have a regulatory penalty in an unregulated environment, but market penalties may be as costly. Thus, a risk-based approach to EU regulatory compliance within the US market will help accelerate deployment where possible while mitigating risk. Other considerations may also include establishing clear legal entities to enable regulatory compliance within different jurisdictions, and building architectural flexibility into AI systems to accommodate those jurisdictions as well as future regulatory changes. Businesses must continuously monitor the performance of their models, maintain awareness of regulatory developments, and be prepared for third-country developments which may follow the EU or US model, or go their own way. [7].
The tension between the prioritization of precaution and human rights versus the championing of innovation and competitiveness is inevitable, and not particularly novel. In this case, however, the relentless, rapid pace of AI development and the resulting capabilities will always outstrip the speed by which policy can adapt. Companies must adopt a hybrid strategy to incorporate robust risk management while remaining agile enough to capitalize on new opportunities. Engagement of all stakeholders will be crucial to shape policies which safely but profitably manage the tension in responsible AI growth.
References
- IDC. 2024. “IDC: Artificial Intelligence Will Contribute $19.9 Trillion to the Global Economy through 2030 and Drive 3.5% of Global GDP in 2030.” IDC Global. https://my.idc.com/getdoc.jsp?containerId=prUS52600524.
- Luget, Angela, Gabriel M. Asaftei, Roger Roberts, Brittany Presten, and Katherine Ottenbreit. 2025. “Insights on responsible AI from the Global AI Trust Maturity Survey.” McKinsey. https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/tech-forward/insights-on-responsible-ai-from-the-global-ai-trust-maturity-survey.
- “America’s AI Action Plan.” 2025. The White House. https://www.whitehouse.gov/wp-content/uploads/2025/07/Americas-AI-Action-Plan.pdf.
- EU. 2024. “EU Artificial Intelligence Act.” EU Artificial Intelligence Act | Up-to-date developments and analyses of the EU AI Act. https://artificialintelligenceact.eu/.
- Dominguez, Maria R. 2025. “EU AI Act: one year on, new measures enter effect.” https://www.cio.com/article/4032894/analysis-of-the-european-ai-regulation-one-year-after-its-entry-into-force.html.
- MacCarthy, Mark. 2025. “How different states are approaching AI | Brookings.” Brookings Institution. https://www.brookings.edu/articles/how-different-states-are-approaching-ai/.
- Dorwart, Hunter, Harry Qu, Tobias Bräutigam, and James Gong. 2025. “Preparing for compliance: Key differences between EU, Chinese AI regulations.” IAPP. https://iapp.org/news/a/preparing-for-compliance-key-differences-between-eu-chinese-ai-regulations.