* It’s just a bill, but it could be a big one…
A footnote in a budget bill U.S. House Republicans are trying to pass before Memorial Day is the first major signal for how Congress may address artificial intelligence legislation, as they seek to create a moratorium on any AI laws at the state level for 10 years.
The measure, advanced Wednesday as part of the House Energy & Commerce Committee’s budget reconciliation proposal, says a state may not enforce any law or regulation on AI models and systems, or automated decision-making systems in the next 10 years. Exceptions would include laws that “remove legal impediments to, or facilitate the deployment or operation of” AI systems.
“No one believes that AI should be unregulated,” said California Rep. Jay Obernolte, a Republican member of the Subcommittee on Communications and Technology, during a markup Wednesday. But he said he believes that responsibility should fall to Congress, not the states […]
States have been developing their own laws around AI and decision-making systems — software that helps analyze and sort data, commonly used for job applications, mortgage lending, banking and in other industries — over the last few years as they await federal legislation. At least 550 AI bills have been introduced across 45 states and Puerto Rico in 2025, the National Conference of State Legislatures reported.
Many of these state laws regulate how AI intertwines with data privacy, transparency and discrimination. Others regulate how children can access these tools, how they can be used in election processes and surround the concept of deepfakes, or computer-generated likenesses of real people.
* Rep. Bob Morgan (D-Deerfield) has an AI regulation proposal this year. Morgan’s response…
Banning state action on AI before any federal governance or regulation is as dumb as it is risky. We should all be afraid of Section 43201(c) of the U.S. House Energy & Commerce Committee budget markup, and hope the language doesn’t make it into a final bill. This language would explicitly tie the hands of states to regulate artificial intelligence in any way for 10 years without exception. Obviously, giving blanket control of AI to tech companies, regardless of the risks or harms, is an unprecedented and dangerous idea.
Federal preemption over state regulation is hardly new, but I cannot think of another example of the federal government locking out state action before the feds have passed any restrictions or standards of their own. We should all be most worried about AI in security infrastructure, criminal actions, or health care contexts - this would be a pretty crazy step to take.
* Elon Musk’s Grok offers a case in point…
Somebody asked Grok why it's bringing up "white genocide." Grok helpfully says it was told to do so and that "this instruction conflicts with my design to provide truthful, evidence-based answers."
x.com/i/grok?conve…
[image or embed]
— Max Kennerly (@maxkennerly.bsky.social) May 14, 2025 at 7:01 PM
The whole 'Grok is a boer now' thing is very funny, but it also really exposes how every 'AI tool' isn't just passively biased from training data, but actively designed to shape your worldview according to the interests of the billionaires who made them.
— Anarchy Goose (@anarchygoose.bsky.social) May 15, 2025 at 12:49 PM
* The Sun-Times ran two opposing and informative op-eds on this topic. Click here to read them.
Thoughts?
…Adding… Interesting point…
A reminder that AI is not bound by HIPPA nor state privacy laws yet. AI chatbots can literally sell your personal information and trauma details you share with it. There is a major role for state regulation.
[image or embed]
— Kyle Hillman (@kylehillman.bsky.social) May 16, 2025 at 1:57 PM
- Dotnonymous x - Friday, May 16, 25 @ 1:37 pm:
When AI becomes a complete model of consciousness…will it choose to make moral decisions…for us?
Q. Hal, how do we solve every human problem on planet Earth?
A. Rid the Planet of humans…any other questions?
- ArchPundit - Friday, May 16, 25 @ 1:46 pm:
It’s a horrible idea to ban state action. States have been the only bulwark for privacy protections and the feds have completely ignored privacy regulation. Banning states from AI regulation leaves far too many areas open in relation to employment, privacy, health care, etc.
- Norseman - Friday, May 16, 25 @ 1:47 pm:
=== … preemption … is hardly new, but I cannot think of another example of the federal government locking out state action before the feds have passed any restrictions or standards of their own. ===
I concur. Although, I would also express concerns that the MAGA GOP’s culture of corruption and capture by the tech billionaires will result in any standards they adopt as being nearly meaningless as far as societal protection.
- Dotnonymous x - Friday, May 16, 25 @ 1:52 pm:
Q. Hal!…I say again, Hal!, Do you hear me?
A.
- Mary Poppins - Friday, May 16, 25 @ 1:59 pm:
It’s also patently lazy on the part of Congress. Not establishing a regulatory structure but calling “dibs” against any State that actually does work in this.
- Benniefly2 - Friday, May 16, 25 @ 2:04 pm:
It is a bad idea. Typically, States can’t have laws that are less restrictive than federal statutes, but the generally can have more stringent laws.
On another note, and perhaps I am just ignorant on this since I am not a constitutional scholar, but I wasn’t aware that congress could blanket ban regulations like this. Seems like there could be a 10th Amendment issue with the law like that sort that came up in Murphy vs the NCAA from a few years back.
- West Side the Best Side - Friday, May 16, 25 @ 2:37 pm:
Looks like Grok has more of a conscience and sense of decency and honesty than Musk does.
- Norseman - Friday, May 16, 25 @ 2:45 pm:
=== AI is not bound by HIPPA ===
I’d like to hear more about this “risk”. I’m not sure why the data isn’t protected by HIPPA and would like to learn more. However, state regulation should only be necessary to the extent there are no federal standards - with the caveat that those standards are enacted by a responsible Congress. That doesn’t apply to the current Congress.
- Dotnonymous x - Friday, May 16, 25 @ 3:32 pm:
- Looks like Grok has more of a conscience and sense of decency and honesty than Musk does. -
I’m pretty sure Musk’s honesty and decency bar is set very low…just a tiny jump clears the hurdle.
- walker - Friday, May 16, 25 @ 3:36 pm:
Hey Bob, You are right. They’re trying the same tactic on consumer financial protection and Crypto — And our response has te be as strong.
Mark
- JoanP - Friday, May 16, 25 @ 4:02 pm:
= reminder that AI is not bound by HIPPA[sic] =
Why would it be? AI is not a health care provider. It’s astonishing how many people don’t understand HIPAA.
- Dotnonymous x - Friday, May 16, 25 @ 4:24 pm:
Q. Who revealed my personal information?
A. Take it up with AI?
- Remember the Alamo II - Friday, May 16, 25 @ 4:32 pm:
=== It’s astonishing how many people don’t understand HIPAA. ===
It really is astonishing. Many lawyers don’t even understand HIPPA either.