Who should control powerful AI?
Which rules can address serious risks without making access depend on a few companies or governments?
What’s at stake
The argument is not only how much regulation AI needs. It is where authority sits, what triggers intervention and who can afford compliance. A rule can address a real harm while also changing competition. That makes the structure of oversight as important as its stated purpose.
The ideas that connect
A national standard changes who can compete
The case against a state-by-state patchwork starts with compliance costs. A large lab can employ teams to navigate different rules; a startup may struggle before reaching customers. The same concern appears in proposals for a federal framework and criticism of escalating state requirements. But uniformity only answers where rules come from. It does not decide what protections the national standard should contain.
Supporting passages
2026-01-23
And what's most important about this debate, which I think a lot of people sometimes don't sometimes miss, is the patchwork is actually most detrimental to early stage young companies and entrepreneurs. If you want to develop a new AI technology, if you want to build something on top of one of our great frontier models, having to figure out how to navigate 50 different rules across 50 different states creates a lot of friction. Ultimately, the big guys are the ones that can succeed in that environment the best. We're spending a lot of time trying to think about how can you create a legislative proposal that can actually deliver on a sensible national framework to solve this regulatory issue.
Open the source and seek to the timestamp above.
2026-07-18
And there was an article in Politico just the other day, it's called Inside Anthropic State-by-State Plan to Ratchet Up AI Rules. And what it says is, quote, AI giant Anthropic is pursuing a strategy of one upmanship that encourages states to impose increasingly tougher AI guardrails rather than align around a single set of regulations. So the basic idea is that they get a set of regulations passed in one state, like California's SB 53 And that was then supposed to be the model, at least for all the blue states. But then with each new state, they actually make the regulations more and more strict, more and more all-encompassing. So there's not actually a stable equilibrium. What they're trying to do is drive each incremental state more and more regulations.
Open the source and seek to the timestamp above.
Punishing harmful conduct differs from approving models
One approach applies existing liability to harmful actions. Another gives an authority a role before a model reaches users. The first leaves more freedom to build but acts after a violation; the second can intervene earlier but controls access to general-purpose tools. The recurring concern about centralized power is that safety review could expand into decisions about which capabilities or viewpoints are acceptable.
Supporting passages
2025-10-03
It's already illegal to conduct a cyber attack. And if you manage to take an AI model and use it as a tool to perform a cyber attack, that's still going to be illegal. Same thing in Colorado, okay? They've got this bill that they want to outlaw algorithmic discrimination, but discrimination is already a violation of the law. So what they're doing there is they're not just going after the business that's performing discrimination. That's already illegal. What they want to do is get into the tool itself, right? And they want to make the developer liable. If their model creates an output that supposedly ends up creating a disparate impact in a decision.
Open the source and seek to the timestamp above.
2026-05-29
Well, I very much agree with the Pope that the biggest risk of AI is the centralization of power and then its misuse against us in some Orwellian way. I think it's government that's going to do that. Not necessarily an individual actor because it's governments that ultimately have the power. So I do worry about the potential for AI to be used to surveil us, censor us, control us, as Orwell described in 1984 So if that's where the Pope is going with this, I very much agree with him. Maybe where we end up in different places is he thinks that government regulation is the way to prevent this. And I would just say that we have to be careful not to empower government too much because if you give government the power to regulate or approve AI development, if you create, say, an FDA for AI, as many people are calling on, that will give government the power to approve models and therefore give notes to model developers.
Open the source and seek to the timestamp above.
Industry coordination can mean very different things
Shared standards, research and incident learning are not the same as mandatory pre-release approval. The debate over industry-led bodies turns on that boundary: can coordination improve practices without becoming a government-backed gatekeeper? Combined with the patchwork concern, the risk is that large incumbents help write rules that smaller competitors must then satisfy. The proposed body’s powers matter more than its reassuring label.
Supporting passages
2026-08-21
And what the government said is, get your act together, guys, or we're going to have to do it for you. And what the MPAA did then was promote standards that then forestalled more intrusive, heavy-handed government action. So that is basically the model that I said I could get behind. But is that the model that Dario supports? No. Dario and I think Demis wants a FINRA, which is basically this rule-making regulatory agency that would not only promulgate standards, but also do pre-release model testing, okay? And it would then report into the government and then officially be enshrined in law. And the reason why everyone says, well, Elon's behind her, I've talked to Elon about this. Elon does not support a FINRA for AI. What he could get behind is an MPAA for AI.
Open the source and seek to the timestamp above.
2026-01-23
And what's most important about this debate, which I think a lot of people sometimes don't sometimes miss, is the patchwork is actually most detrimental to early stage young companies and entrepreneurs. If you want to develop a new AI technology, if you want to build something on top of one of our great frontier models, having to figure out how to navigate 50 different rules across 50 different states creates a lot of friction. Ultimately, the big guys are the ones that can succeed in that environment the best. We're spending a lot of time trying to think about how can you create a legislative proposal that can actually deliver on a sensible national framework to solve this regulatory issue.
Open the source and seek to the timestamp above.
What remains unresolved
Ask what harm a rule addresses, who enforces it, how decisions can be challenged and whether new entrants can comply. Neither voluntary coordination nor a national framework resolves those questions by name alone.
A selected guide · September 2025–September 4, 2026 · Sources and method