White House backs 'light touch' AI
The White House released a legislative blueprint urging Congress to take a “light touch” approach to AI regulation, focusing federal guardrails on fraud, safety and speech rather than sweeping new controls. The plan leaves key workforce and social‑impact risks unresolved and sits uneasily with state action — California, Colorado, Texas and Utah have already passed their own AI statutes. (boston.com) (govinfosecurity.com)
The White House published a four-page legislative document titled “A National Policy Framework for Artificial Intelligence” on March 20, 2026. (natlawreview.com) The framework follows President Trump’s December 11, 2025 executive order “Ensuring a National Policy Framework for Artificial Intelligence,” which directed administration offices to prepare legislative recommendations for Congress. (whitehouse.gov) The document asks Congress to codify the administration’s Ratepayer Protection Pledge and to streamline federal permitting so data centers can use on‑site, behind‑the‑meter power generation to accelerate infrastructure buildout. (whitehouse.gov) The White House notes state legislatures have introduced more than 1,000 AI bills in recent cycles and frames a federal law as a response to that patchwork of state activity. (whitehouse.gov) Senator Marsha Blackburn released a nearly 300‑page discussion draft called the “Trump America AI Act” on March 18, 2026, aiming to translate the administration’s executive‑order priorities into comprehensive statutory text. (rollcall.com) Seven major tech firms — Amazon, Google, Meta, Microsoft, OpenAI, Oracle and xAI — signed the administration’s Ratepayer Protection Pledge in early March committing to fund new data‑center energy and infrastructure costs. (perkinscoie.com) The White House blueprint specifically recommends commercially reasonable age‑assurance measures such as parental attestation and urges Congress to avoid ambiguous content standards that could spawn broad litigation. (whitehouse.gov) Legal advisors and law firms flagged that the framework leaves open whether model training on copyrighted material violates copyright or qualifies as fair use, and the proposal does not create a new federal AI regulator. (sullcrom.com)
