Canadian here. Generative AI is available around the world and with most of the foundational LLMs being American, this could have global ramifications.
We've already seen Big Tech chafe at EU privacy laws. Such a decision in America could embolden them and even the American government to go after regulators in other countries. If they get carte blanche at home, they could become increasingly ungovernable everywhere else.
The bill cannot pass with such language for this reason and all the reasons you stated in this letter.
I was gonna say the same but didn’t wanna turn this into some he-said-she-said slog. It’s always astounding to me where conservative reps will draw the line when it comes states rights, going so far as to “fight” for them long as they clearly don’t interfere with corporate entities’ abilities to make some fat wads of cash.
Good luck with this effort. This curb on state's regulatory powers, vis a vis the AI industry, is very important. Galvanizing the citizenry to safeguard their interests is not an easy task. The pacification of America has progressed far. Let us hope it has not atrophied altogether.
We are racing into policy decisions about this tech when we barely understand what the systemic effects of the policy will be - understanding of the broader social consequences usually lags behind tech adoption, an observation made years ago by Astro Teller (diagram from Thomas Friedman’s book “Thank You For Being Late”).
“Move fast and break things” is OK for private industry where you own the risks, but it shouldn’t apply to our kids, our students, or medical care, or in many other areas that we haven’t fully thought through the potential second and third order consequences yet. We need to be specific about who we’ll be placing the burdens of the risk upon if we make big decisions too hastily in search of short term gains.
One of the issues I see with trying to control/align AI in a market environment is that it will inevitably lead to AIs being trained to lie to regulators. They will be trained on Machiavellian data!
What these people obsessed with the "race" think, when it is not just a money grab, is that one day, there will be similar models from the USA and China. The next day, one of the models will have an advantage that will be so overwhelming as to enable it to consistently and reliably outwit not only the other model, but all human beings and existing AI combined, never mind the probabilistic output, which the country that produces it will use to quite literally Take Over the World.
So if China reaches it first, the world will eternally be under the rule of Xi Jinping Thought, whereas if the USA does, it will forever be a capitalist paradise governed by the invisible hand of the market. Or thousand-year MAGA, take your pick. Needless to say, this singularitarian—not just singularitarian, but radically singularitarian—scenario is just one possiblity.
Not even necessarily a particularly probable one to my judgement: if a chess engine with an Elo score of 3501 will not consistently outplay one as 3500, why assume that a slightly better AI automatically means world domination relative to a slightly worse one? But this does appear to be more or less what they think.
This is a dangerous attempt at government overreach on behalf of Silicon Valley. It's the type of corruption and regulatory capture that should be opposed by every US state and citizen, and as a Canadian you all have my support in this fight against regulatory capture by the elites of tech industry.
More Federal overreach. Exploitive, intrusive, dangerous.
I can't escape the suspicion that the goal is compliant, passive employees and consumers who won't demand their rights. Who will simply accept the abuse and hope for the best.
We're watching this process unfold from the EU with real concern. And we're firmly determined to avoid the mistakes and abuses we see in the US.
The US can't be relied on to regulate AI responsibly—especially when efforts like this aim to strip states of their ability to respond to real harms already unfolding. Blocking local legislation doesn't create consistency, it creates a vacuum—one that favors centralized power and unchecked extraction.
If national governments won't act in the public interest, then it's time for international accountability. Sanctions should be applied to corporations and individuals—executives, board members, and investors—who are aggressively deploying destabilizing AI systems without meaningful oversight. Tariffs on extractive platforms should be coordinated globally.
And the International Criminal Court must open whistleblower protections for AI researchers. These labs are no longer just research hubs. They are de facto weapons manufacturers, operating without constraint, shaping the future at planetary scale with no democratic mandate.
If the US chooses corporate impunity, the world must choose restraint—for everyone’s sake.
As you point out, the US Congress has ignored the many pleas for a real privacy bill. This is what to expect when the reps do the bidding of their [corporate] donors and not their voters. In the UK, we saw the last Tory PM cuddle up to Musk and AI, and there are rumors that the Labour Party might trade the Digital bill and associated legislation for a larger trade agreement with the US. OTOH, the EU and other European/Scandinavian countries are looking to avoid US company software and platforms and build their versions. Musk is not happy with the EU building its comsat swarms to allow it to back away from Starlink, and of course, he is principally responsible for the collapse of Tesla car sales there. Powerful near-monopoly platforms need to be broken up, especially to prevent unfair competition and self-dealing. We need concerted "trust-busting" again in the US.
The AI overlords are intent on doing as they please, and do not regard themselves as subject to regulation - unless they make those regulations. We really need regulations at the federal level that prevents the goal of replacing humans in the economy by machines, rather than as tools to aid humans. What they seem to fail to understand is that unless there is healthy consumer demand, there is little economic demand.
The separation of powers called. They want to remind everyone that there is a big difference between the courts applying the dormant commerce clause and Congress overstepping its authority.
Seems like trenches are being dug. Perhaps values might be a way forward that allows those operating in values-based systems within values-based jurisdictions to differentiate themselves from those who are not? And enough room for some differences to continue to be explored?
Canadian here. Generative AI is available around the world and with most of the foundational LLMs being American, this could have global ramifications.
We've already seen Big Tech chafe at EU privacy laws. Such a decision in America could embolden them and even the American government to go after regulators in other countries. If they get carte blanche at home, they could become increasingly ungovernable everywhere else.
The bill cannot pass with such language for this reason and all the reasons you stated in this letter.
It could also deincentivize other governments from drafting meaningful regulations.
The hypocrisy for a Congress that is majority Republican to rush through a bill that takes away state rights is astounding.
I was gonna say the same but didn’t wanna turn this into some he-said-she-said slog. It’s always astounding to me where conservative reps will draw the line when it comes states rights, going so far as to “fight” for them long as they clearly don’t interfere with corporate entities’ abilities to make some fat wads of cash.
I object in principle to the notion of passing laws that you can't pass laws.
Good luck with this effort. This curb on state's regulatory powers, vis a vis the AI industry, is very important. Galvanizing the citizenry to safeguard their interests is not an easy task. The pacification of America has progressed far. Let us hope it has not atrophied altogether.
We are racing into policy decisions about this tech when we barely understand what the systemic effects of the policy will be - understanding of the broader social consequences usually lags behind tech adoption, an observation made years ago by Astro Teller (diagram from Thomas Friedman’s book “Thank You For Being Late”).
“Move fast and break things” is OK for private industry where you own the risks, but it shouldn’t apply to our kids, our students, or medical care, or in many other areas that we haven’t fully thought through the potential second and third order consequences yet. We need to be specific about who we’ll be placing the burdens of the risk upon if we make big decisions too hastily in search of short term gains.
One of the issues I see with trying to control/align AI in a market environment is that it will inevitably lead to AIs being trained to lie to regulators. They will be trained on Machiavellian data!
What these people obsessed with the "race" think, when it is not just a money grab, is that one day, there will be similar models from the USA and China. The next day, one of the models will have an advantage that will be so overwhelming as to enable it to consistently and reliably outwit not only the other model, but all human beings and existing AI combined, never mind the probabilistic output, which the country that produces it will use to quite literally Take Over the World.
So if China reaches it first, the world will eternally be under the rule of Xi Jinping Thought, whereas if the USA does, it will forever be a capitalist paradise governed by the invisible hand of the market. Or thousand-year MAGA, take your pick. Needless to say, this singularitarian—not just singularitarian, but radically singularitarian—scenario is just one possiblity.
Not even necessarily a particularly probable one to my judgement: if a chess engine with an Elo score of 3501 will not consistently outplay one as 3500, why assume that a slightly better AI automatically means world domination relative to a slightly worse one? But this does appear to be more or less what they think.
This is a dangerous attempt at government overreach on behalf of Silicon Valley. It's the type of corruption and regulatory capture that should be opposed by every US state and citizen, and as a Canadian you all have my support in this fight against regulatory capture by the elites of tech industry.
I hope our new AI minister Evan Solomon doesn't take a similar hands-off approach. I plan to write to him and tell him what's at stake.
More Federal overreach. Exploitive, intrusive, dangerous.
I can't escape the suspicion that the goal is compliant, passive employees and consumers who won't demand their rights. Who will simply accept the abuse and hope for the best.
We're watching this process unfold from the EU with real concern. And we're firmly determined to avoid the mistakes and abuses we see in the US.
The US can't be relied on to regulate AI responsibly—especially when efforts like this aim to strip states of their ability to respond to real harms already unfolding. Blocking local legislation doesn't create consistency, it creates a vacuum—one that favors centralized power and unchecked extraction.
If national governments won't act in the public interest, then it's time for international accountability. Sanctions should be applied to corporations and individuals—executives, board members, and investors—who are aggressively deploying destabilizing AI systems without meaningful oversight. Tariffs on extractive platforms should be coordinated globally.
And the International Criminal Court must open whistleblower protections for AI researchers. These labs are no longer just research hubs. They are de facto weapons manufacturers, operating without constraint, shaping the future at planetary scale with no democratic mandate.
If the US chooses corporate impunity, the world must choose restraint—for everyone’s sake.
Sad that someone who rightfully questions the authority of AI isn't doing the same with government.
As you point out, the US Congress has ignored the many pleas for a real privacy bill. This is what to expect when the reps do the bidding of their [corporate] donors and not their voters. In the UK, we saw the last Tory PM cuddle up to Musk and AI, and there are rumors that the Labour Party might trade the Digital bill and associated legislation for a larger trade agreement with the US. OTOH, the EU and other European/Scandinavian countries are looking to avoid US company software and platforms and build their versions. Musk is not happy with the EU building its comsat swarms to allow it to back away from Starlink, and of course, he is principally responsible for the collapse of Tesla car sales there. Powerful near-monopoly platforms need to be broken up, especially to prevent unfair competition and self-dealing. We need concerted "trust-busting" again in the US.
The AI overlords are intent on doing as they please, and do not regard themselves as subject to regulation - unless they make those regulations. We really need regulations at the federal level that prevents the goal of replacing humans in the economy by machines, rather than as tools to aid humans. What they seem to fail to understand is that unless there is healthy consumer demand, there is little economic demand.
The separation of powers called. They want to remind everyone that there is a big difference between the courts applying the dormant commerce clause and Congress overstepping its authority.
There’s been a whole lot of overstepping going round in DC these past few months.
Seems like trenches are being dug. Perhaps values might be a way forward that allows those operating in values-based systems within values-based jurisdictions to differentiate themselves from those who are not? And enough room for some differences to continue to be explored?