White House AI Framework Protects AI Companies, Not People

Today, the Trump Administration put out its recommendations for a national policy framework on artificial intelligence (hereinafter, the Framework). The Framework sets forth some elements that the Administration would like legislators to incorporate into any AI proposals at the national level and builds off of December’s Executive Order on AI but fails to address any of AI’s most pressing harms.

EPIC Executive Director and President Alan Butler said, “The White House’s framework would promote corporate interests over the public interest. It is light on protection and heavy on promotion of dangerous AI systems. The American people deserve better, and Congress should do better than this.”

This Framework is the latest in the Trump Administration’s continued (failed) efforts to preempt the AI protections that states have been putting in place for their residents in the face of federal inaction. As EPIC pointed out when the Executive Order was released, the White House simply does not have the power to dictate legislative action. The same goes for the Framework. The Framework, on its own, does not change existing federal or state laws that regulate uses of AI.

Lawmakers looking to the Framework for inspiration on how to address the numerous active harms stemming from AI will be disappointed. The policy recommendations contained in the framework are vague and conceptual. The strength of any legislative proposals on AI threats to children, privacy, the economy, and individual rights resulting from this Framework will depend entirely on the details.

What we are able to glean from the Framework is not promising. The Framework promotes nearly unfettered AI development and use—including open instructions to preempt state laws that may “act contrary to the United States’ national strategy to achieve global AI dominance,” directives to incorporate AI training into education and workforce training, a mandate to make federal datasets available for AI training, and suggestions that AI infrastructure permitting should be “streamlined.” While the Framework does acknowledge some carveouts for generally applicable existing consumer protection laws and laws that protect children, these carveouts lack reference to state privacy or AI laws being passed in response to new and acute risks created by AI systems.

The Framework fails to address the full scope of ways AI harms people, instead focusing on pushing the country forward in AI adoption, regardless of the human costs. There is no mention of general privacy protections or AI’s use of personal data, but an entire section framed in the guise of protecting free speech but which, in practice and in light of the Administration’s public stance against certain speech, is meant to ensure that AI companies carry conservative viewpoints. The Framework urges Congress to leave intellectual property disputes to the courts while noting that the Administration believes that using copyrighted material to train AI models is not a copyright violation. The vague suggestions for protections for children are both unclear and inadequate to address AI harms and are more focused on limiting substantive obligations on AI companies to avoid “excessive litigation.”

Once again, the White House is promoting the interests of AI companies over those of the American people and attempting to encroach further on powers designated to the legislature. We strongly urge U.S. lawmakers to push for thoughtful AI regulation that focuses on safety, transparency, and protecting human rights—not this AI company wish list.

EPIC has long worked to draw attention to the serious and ongoing risks of AI, including through reports, model bills on chatbots, state testimony, and more. Our work on AI is people- and privacy-centered, pushing for responsible AI use and development.