U.S. lawmakers are drafting new artificial intelligence legislation while facing criticism that they lack a fundamental understanding of the technology.

This knowledge gap is critical because policy decisions regarding AI safety, economic impact, and regulation may be based on flawed technical assumptions. If legislators cannot grasp how these systems operate, the resulting laws could be ineffective or counterproductive.

On June 2, 2026 [1], bipartisan legislation was announced that aims to restrict the ability of individual states to regulate AI. This move suggests a push for a unified federal framework to manage the technology's rapid growth.

However, critics argue that the legislative process is flawed. An editor from Gizmodo said that lawmakers are still playing catch-up with the technology, and many seem to be drafting policy without a basic grasp of how AI works [1].

Specific proposals have also drawn fire. Sen. Bernie Sanders (I-VT) proposed an AI wealth fund, but a contributor from Reason said that Senator Sanders lacks a basic understanding of AI, as evidenced by the assumptions in his wealth-fund proposal [2].

Other members of Congress have engaged with the topic through formal testimony. Sen. Marco Rubio (R-FL) testified before Congress on June 2, 2026 [3], marking his first appearance before the body since the start of the Iran war.

Despite these hearings and the introduction of bills, some industry observers believe the government is lagging. TechCrunch staff said that even as the worst hacks of 2026 unfold, Congress remains behind on AI policy [4].

The tension between the need for rapid regulation and the lack of technical expertise among policymakers continues to shape the debate on Capitol Hill.

“Lawmakers are still playing catch-up with the technology.”

The disconnect between technical AI evolution and legislative literacy creates a regulatory vacuum. While bipartisan bills indicate a desire for federal oversight, the reliance on non-expert policymakers may lead to legislation that focuses on perceived risks rather than the actual mechanical functions of large-scale AI models.