Policymakers cannot effectively regulate artificial intelligence without developing a deeper understanding of how the technology actually works, according to UN officials and AI governance experts. The challenge isn't turning legislators into engineers, but rather ensuring they grasp enough about AI systems to ask critical questions, recognize misleading corporate claims, and translate values like transparency and fairness into enforceable laws.

What's Blocking Effective AI Regulation in the U.S.?

The United States currently lacks a single federal rulebook for AI governance. Instead, oversight is scattered across a patchwork of state laws, voluntary industry standards, existing consumer protection rules, and lawsuits filed after harm occurs. This fragmented approach means that while AI systems are already influencing decisions about healthcare, hiring, pricing, and insurance claims, most of these applications operate without comprehensive federal scrutiny.

Three states have passed laws focused on frontier AI safety and transparency: California, New York, and Illinois. Meanwhile, Colorado and Texas have taken different regulatory approaches. However, these state-level efforts have significant limitations. Many focus on mitigating catastrophic risks from the largest AI systems, which exempts most AI applications already in deployment from meaningful oversight. This means the harms people experience today, such as denied insurance claims or discriminatory hiring decisions, receive far less attention than hypothetical future risks.

"We do not need to turn policymakers into engineers. Still, they need the fluency to ask the right questions, to recognise when a company's claims don't hold up, and to understand values like transparency, fairness, and accountability well enough to write them into law in a way that's actually enforceable," said Prof. Tshilidzi Marwala, Rector of the United Nations University.

Prof. Tshilidzi Marwala, Rector, United Nations University

How Does Europe's Approach Differ From the U.S. Model?

The European Union has taken a fundamentally different path by passing the AI Act, a comprehensive regulatory framework that categorizes AI uses by risk level. The framework identifies low-risk applications that require minimal scrutiny, higher-risk domains that demand significant attention, and certain uses deemed too dangerous to permit at all. This risk-based approach allows regulators to tailor oversight to specific contexts and use cases.

However, the European model also faces challenges. Much AI is developed as general-purpose technology rather than for specific applications, which means it requires scrutiny before widespread deployment, not after problems emerge. The U.S. has largely rejected this broad regulatory approach, though some policymakers have considered adopting similar frameworks.

What Role Are U.S. Agencies Playing in AI Governance?

In the absence of comprehensive federal legislation, several U.S. agencies have stepped into the governance vacuum. The National Institute of Standards and Technology (NIST) created the AI Risk Management Framework, which provides voluntary guidance for companies managing AI risks. However, these guidelines are not binding and lack enforcement mechanisms, meaning companies can essentially grade their own homework without meaningful consequences.

The Federal Trade Commission (FTC) and Consumer Financial Protection Bureau (CFPB) also play roles in AI oversight, but their authority is limited to enforcing existing consumer protection laws rather than establishing AI-specific regulations. This reactive approach means regulators typically intervene only after AI systems cause documented harm, rather than preventing problems proactively.

How Can Policymakers Build Better AI Literacy?

• Technical Fluency Development: Legislators need sufficient understanding of AI systems to evaluate corporate claims critically and recognize when companies overstate capabilities or downplay risks without becoming engineers themselves.
• Values-Based Regulation: Policymakers must grasp core AI governance principles like transparency, fairness, and accountability deeply enough to translate them into legally enforceable standards that actually constrain corporate behavior.
• Behavioral Change Strategy: Effective regulation requires building understanding among legislators, companies, and the public simultaneously, since changing behavior across all stakeholders depends on shared comprehension of AI's actual capabilities and limitations.

The knowledge gap extends beyond policymakers. Educational institutions are also grappling with how AI will reshape learning itself. Universities traditionally assess students on analytical work that AI systems can now perform, forcing educators to rethink not just teaching methods but the fundamental purpose of assessment. This institutional challenge mirrors the broader governance problem: understanding must precede effective response.

The stakes are substantial. AI systems already shape consequential life outcomes through healthcare decisions, employment screening, and financial services. Yet most people affected by these systems have no visibility into how decisions are made or opportunity to challenge them before harm occurs. Without policymakers who understand AI deeply enough to write enforceable protections, this accountability gap will likely widen as AI capabilities expand.

The path forward requires simultaneous action on multiple fronts: building regulatory literacy among policymakers, establishing binding standards rather than relying on voluntary compliance, and moving from reactive enforcement to proactive oversight of high-impact AI applications already in deployment.