In the episode of Consumer Finance Monitor Podcast being released today, we explore the White House's National Policy Framework for Artificial Intelligence published on March 20, 2026. This new framework represents the Administration's most concrete attempt yet to shape the future of AI governance in the United States. While it does not carry the force of law, it offers a revealing look at the policy direction the Administration hopes Congress will take.

Joining our host, Alan Kaplinsky (founder, chair for 25 years and now Senior Counsel of the Consumer Financial Services Group), for this discussion were Charlie Bullock (Senior Research Fellow at The Institute for Law and AI), Kristian Stout (Director of Innovation Policy at the International Center for Law & Economics), and Greg Szewczyk, head of Ballard Spahr's Privacy and Data Security Group. Below are the key takeaways from the conversation.

From Principles to Policy: A Clear Shift

One of the most striking aspects of the new framework is how sharply it departs from last year's more principles-based "White House AI Action Plan." That earlier effort emphasized risk awareness, governance principles, and a balanced approach to innovation and regulation. On October 30, 2025, we produced a webinar entitled: "AI in Financial Services: Understanding the White House Action Plan – and What It Leaves Out", which featured the same speakers as the podcast being released today, plus Dean Ball, former White House senior advisor and one of the architects of the White House AI Action Plan. This webinar was then re-purposed into a two-part podcast series released on December 4 and 10, 2025.

By contrast, the new framework is short, just a few pages, light on detailed policy prescriptions, and heavily focused on limiting regulation, particularly at the state level.

As Charlie Bullock observed, the document is notable as much for what it doesn't include as for what it does. Rather than proposing robust federal oversight, it largely outlines areas where the government should refrain from acting.

Federal Preemption Takes Center Stage

The framework's most consequential and controversial feature is its strong endorsement of federal preemption of state AI laws.

It proposes broad preemption in areas such as:

· AI development

· Liability for third-party misuse of AI systems

· Restrictions on AI-enabled activities that would otherwise be lawful

At the same time, it preserves certain state authorities, including:

· Zoning and infrastructure decisions

· State use of AI

· "Generally applicable" laws (e.g., fraud, consumer protection, and child safety)

This raises a critical question: How meaningful are these carve-outs? As we discussed, broadly worded exceptions, particularly for state "police powers", could significantly limit the practical reach of federal preemption and potentially preserve a patchwork of state regulation.

The Patchwork Problem Isn't Going Away

Even with federal action, the reality is that state-level AI regulation is already underway. Laws like Colorado's AI Act and emerging chatbot regulations illustrate how quickly states are moving.

Greg Szewczyk noted that, unlike privacy law, where states have largely converged around similar frameworks, AI regulation could diverge in more fundamental ways. Without a consistent federal baseline, companies may face:

· Increased compliance costs

· Operational complexity

· Uncertainty in deploying AI tools across jurisdictions

Interestingly, some state regulators (including Democrats) may ultimately favor a well-crafted federal preemption regime if it provides clarity without sacrificing core protections.

Innovation First—But Who Benefits?

The framework strongly emphasizes:

· AI infrastructure buildout

· Faster permitting

· Regulatory sandboxes

· Access to federal datasets

Kristian Stout highlighted that these priorities could accelerate innovation but they are not automatically startup-friendly. Large incumbents may benefit disproportionately due to:

· Greater access to compute resources

· Established compliance capabilities

· Ability to absorb regulatory costs

This tension between promoting innovation and preserving competition remains unresolved.

Child Safety, IP, and Free Speech: More Questions Than Answers

The framework touches on several critical areas but leaves key details unsettled:

Child Protection

It endorses tools like age verification and parental controls but offers little guidance on implementation. Compared to proposals like the Kids Online Safety Act (KOSA), the framework appears less aggressive and more preemptive of state innovation.

Intellectual Property

Rather than legislating, the framework defers to the courts on issues like:

· Fair use in AI training

· Output infringement

This "wait and see" approach avoids premature policymaking but prolongs uncertainty.

Free Speech

A novel component aims to prevent government "jawboning" of AI providers; i.e., informal pressure to shape outputs. While rooted in legitimate First Amendment concerns, its ultimate scope and constitutionality remain unclear.

No New AI Regulator—For Now

The framework rejects the creation of a centralized AI regulator, instead relying on existing agencies.

This approach has clear advantages:

· Agencies already understand their sectors

· Avoids bureaucratic duplication

But it also raises concerns:

· Limited technical expertise

· Resource constraints

· Inconsistent oversight across agencies

As discussed, a hybrid model, combining agency expertise with centralized technical guidance, may ultimately emerge.

Will Anything Actually Pass?

Perhaps the most sobering takeaway: major AI legislation is unlikely in the near term.

As Charlie Bullock put it bluntly, companies should not invest significant resources preparing for this specific framework. The political reality is: