Congressional representatives Jay Obernolte and Lori Trahan introduced a bipartisan proposal Wednesday aimed at regulating artificial intelligence — a rare joint effort in a deeply divided Capitol. The legislation tries to strike a tricky balance: encourage innovation, set federal standards, and leave room for states to chart their own course. But the bill lands in a Congress where even modest tech bills stall for months.
What the proposal covers
The plan lays out a framework for federal oversight of AI systems while explicitly preserving state autonomy. Obernolte and Trahan, who co-chair the bipartisan AI working group, say the goal is to prevent a patchwork of conflicting state laws without wiping out local experimentation. The proposal doesn't prescribe specific rules for every use case — instead it sketches principles that would guide future rulemaking.
The core tension is familiar: too much federal control could smother start‑ups; too little could leave consumers exposed to harmful AI. The bill leans toward a light federal touch, with states free to add their own guardrails as long as they don't contradict national standards.
Why progress has been slow
Despite the bipartisan backing, the proposal faces an uphill slog. Lawmakers on both sides say the sheer complexity of AI governance — from data privacy to liability for autonomous decisions — makes it hard to write laws that don't break things. Several committees claim jurisdiction, further slowing the timeline. Congressional aides familiar with the talks say the working group spent months just agreeing on definitions like what counts as a “high‑risk” AI system.
Obernolte acknowledged the drag in a brief statement after the unveiling, noting that “building consensus on something as far‑reaching as AI takes time.” Trahan added that the bill is meant to start a conversation, not deliver a finished product overnight. Neither offered a timeline for floor votes.
The state autonomy question
California and New York have already passed their own AI disclosure and testing laws. Industry groups have warned that a dozen different state regimes would create compliance chaos. The bipartisan proposal tries to head that off by creating a baseline federal standard without preempting states entirely — a compromise that pleases neither side fully.
Tech companies have been lobbying heavily for a single national rulebook, while civil‑rights groups push for tougher federal requirements. The proposal dodges the hardest decisions, kicking specifics to future rulemaking that would require further legislation or agency action.
The next concrete test comes in two weeks, when the House Energy and Commerce subcommittee on innovation holds a hearing on the bill. Whether it gets a markup — or languishes like dozens of other tech bills — will signal whether bipartisanship on AI can translate into law.
