Can someone tell me why I shouldn't be as worried about this as it sounds? I haven't forgotten the crypto export restriction debacle. Why are we walking back into that garbage fire, but N times over?

I could very well see this as being one of those "will never see the light of day out of a congressional subcomittee" situations but I'm somewhat surprised at the complete lack of discussion on the topic.

I'm not particularly worried for a few reasons: because there are multiple giant companies that will oppose restrictions on AI software. Because with PGP the government backed down instead of pursuing a legal case that could end with the Supreme Court deciding on the First Amendment status of source code. Because the "representative technologies" list published in the Federal Register is clearly speculative; it includes not-there-yet ideas like "smart dust" and "molecular robotics" plus absurdly broad categories like "Systems-on-Chip (SoC)."

I don't think anybody actually likes export control restrictions, but do you have a better idea for how we might keep this technology away from foreign governments? It appears that the only thing worse than export control restrictions is all the alternatives.

I have patents and publications in multiple of the spaces listed under that whitepaper. And honestly? I don't WANT to keep it away from foreign governments. I don't think you could succeed, mind you, none of what I work on is complex enough to avoid a black box reinvention, but in trying to wall off the information transfer you'd see the whole slew of negative externalities as we saw in the aformentioned crypto regulation.

Fundamentally, I've been in no way convinced that these techs would even move the needle sufficiently in expanded capabilities for our enemies relative to the costs and limitations incurred by trying to fight what is, at its core, free spread of information. (Not to mention externalities via increased protectionism and overhead for american employees and practitioners)

Your "only thing worse is the alternative" suggests that the world should have become concretely worse since the rollback of export restrictions in the 2000's, which if you remember the arbitrary and spurious limitations to consumer software under the guise of "protection" prior in the same light I do, certainly didn't happen.

"the restrictions are intended to be for your benefit."

This gives me small comfort; since as much as I'm loathe to quote Reagan, "The most terrifying words in the English language..."

Yes, I am thankful we don't live an an even more authoritarian state. That doesn't mean I won't strive, advocate, and fight for one both less iron fisted and more open than the alternative you propose.

(Bluntly as well, you haven't done much to convince me WHY they should be needed, nor rebut my precedent re: crypto regulation, so an appeal to authority isn't a strong argument; it's quite the reducto-ad-absurdum to compare much of the whitepaper list in the OP to "nuclear secrets" in terms of potential risks, and even moreso, seems to support my stance, in that restriction on nuclear secrets has canonically NOT stopped our enemies)

The world has what, maybe a dozen countries with nuclear weapons after 7 decades? I think the restrictions have worked very very well, especially considering just how desirable nuclear weapons are.

The "why" is pretty simple. Technology leads to the rise and fall of nations. There is a huge benefit to the people of a nation when their nation has a technology advantage. This applies both in war and in peace.

Like it or not, each nation is fiercely competing with every other nation. The relative degree of winning or losing determines the level of prosperity. Technology is unavoidably a part of this. It can't be just set aside. You're part of a team (with your fellow citizens) and you are expected to avoid helping the other teams win.

First: Having nuclear weapons is, for the most part, highly undesirable in a modern geopolitical context, for anyone but a few rogue states. (the ones who fit into the small niche of wanting the tech and having little to lose can trivially achieve it and regularly use it to play the superpowers like a fiddle, see Iran, NKorea, and Pakistan to a lesser extent) In the current world order most small countries would much rather rely on a superpower to "have their back" than attract the undue attention and costs of developing their own nuclear power. (and probably force-of-international-law as well, likely a far more powerful lever in terms of nation state behavior than lack of technology transfer, especially seeing how one can find most of what used to be "nuclear secrets" on the web or via "nation-state-IP-transfer")

"you are expected to avoid helping the other teams win."

This to me describes most of, and most of the darkest, parts and mindsets of human history. Hard as it may seem to believe there's another path, I refuse to resign myself to a nationalist mindset.

And honestly? The short term matters so little to me, outside of selfish concerns. Over the next few thousand years superpowers will come and go entirely regardless of what silly code I write. "winners" and "losers" will be chosen on tectonic scales. But if perhaps a slight bit of increased communication can move human progress forward as a whole over the next millenia? I'll risk much for that goal.

Frankly, I don't see technology as the differentiator you think it is. Overall economic horsepower, yes. Technology, no. Back in the day, the Germans had jets before we did. Better rifles, at the start. Better tanks, too. In the same way, some AI models likely won't make or break a future conflict. The ability to have 300 million people turn their hands to war will.

I'd keep hammering the point as well, that we _have factually loosened crypto restrictions_ and the world has not ended; in fact we've gotten far better consumer crypto utilization that have canonically improved our access to new capabilities, tools and products, and our "great enemies" from the days of restriction aren't suddenly more empowered to talk behind our backs. The whole effort was moot.