On the other hand, we did ban TikTok (which is currently unavailable on the app stores because of the ban).
I can think of few ways to more effectively destroy any US advantage in AI compared with the sheer efficiency of making it illegal to learn from what competing countries have achieved. From the article, it sounds like the proposed legislation is deeply confused about what "downloading DeepSeek" means--they're _talking_ about banning the app, while _writing_ laws to ban "the importation into the United States of artificial intelligence or generative artificial intelligence technology or intellectual proprietary developed or produced in the People’s Republic of China" which is a lot broader. Is it proposing to ban US citizens from reading research papers written by Chinese citizens? Or from publishing research in places that might be read by them? Apparently the EFF is concerned that the language of the bill is, indeed, that broad.
https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...
It is not speculation at all to say that tiktok can both collect data and manipulate populations. Whether they actually do doesn't even matter. The problem is you can't prove that they have not, are not, or will not.
The only problem is 50 other apps all pose essentially the same risks, and tiktok is hardly special.
The contrast is any of the various open source activity pub apps. Any given server could be doing anything, but there is no single black box server that everyone has to live on. There are 1000 different servers and a bunch of open source code that all serve as verification checks on each other. No aberrant server can hide its skewed behavior for long when there are other servers that didn't F with the code and so their behavior matches the public code, and so the aberrant ones can be exposed by their behavior and stats deviating from the norm, even if you can't trust that a given server is really running clean code.
No harm in pulling down R1 right now anyway (which is what I might have done…).
This exactly!
i think the tiktok ban ought to be considered unconstitutional.
Unless proof can be found that tiktok is being used by foreign actors to manipulate votes or some other shady shit (and only via tiktok itself, rather than a broad scheme that uses all social media), tiktok should not be banned. It is against the free speech that the US stands for.
If algorithmic manipulation of voters is considered a threat, then pass a law to social media companies that they must explain the algorithm etc, and audit it. It's exactly what GDPR mandates, in their automated decision making and profiling clauses (https://gdpr-info.eu/art-22-gdpr/). Therefore, this is not without precedent, and is also more effective at regulating social media, not just tiktok.
Therefore, the tiktok ban is performative. Like how the claims for DEI and such are performative.
> yeah TikTok ban was a very bad precedent
The TikTok ban didn't set a precedent, it's following a general precedent. The most obvious precedent is Grindr [0]. What makes TikTok unusual is they refused to sell.
The precedent is also very old. There are usually laws regarding doing business with enemies during wartime, for example. That's the same basic idea: restriction on business/trade due to national security issues. The ones everyone knows about are US car companies and IBMs doing business with the Nazis. Typically, AFAIK, companies are required to divest to a subsidiary, which is what the US wanted ByteDance to do.
At least that's the right conceptual framing IMO, whether or not you think the ban was a bad idea.
CISA considers China the "most active and persistent cyber threat to the U.S. Government, private sector, and critical infrastructure networks" [1]. So while the US isn't technically in a cyber war with China, they basically are.
[0] https://www.theverge.com/2020/3/6/21168079/grindr-sold-chine...
[1] https://www.cisa.gov/topics/cyber-threats-and-advisories/nat...
... That seems _absurdly_ broad; would arguably hit open source software, academic papers, patents... Like, taken at face value, this would mean "no longer do AI stuff in the US _at all_"; it would be impossible to do anything much without risking violating this.
It's unlikely that this will ever get out of committee, of course.
Are you confused by the questions/arguments in it?
You get less jail time for committing 34 felonies.
Lawfare was tried, and it failed. I hope it dies the death it deserves instead of becoming the new normal.
We have a deeper societal perception to fix first before we can even think about a justice system focused on rehabilitation. It'd also be nice to remove that certain clause in the 13th amendment while we're at it.
relevant: https://web.archive.org/web/20130208124604/https://www.popeh...
The moment the government wants to punish someone over this they will grab some random kid that's barely done anything and threaten them with the full 20 years to serve as an example to others.
Oh goodness me!. If that's true, then why not just completely block China's IP addresses from our Internet? Why not just block all dollars from entering China?
Oh right, because that's not the truth. It's certainly not the whole of it anyway.
Total U.S.-China trade is roughly 760 billion dollars and includes some of the most wealthy and best connected companies in the U.S. Dollars of wealth and investments created through trade with Chinese companies (many of which are connected right to the CCP) pervade various levels of American society and the finances of who knows how many high.level politicians. The exact same thing applies to Chinese society and politicians with money gained from Chinese trade with American companies. That's how large.scale trade just works between two societies, and it's usually a good thing because it's better to trade goods and services than it is to trade bullets and missiles..
For his incoherent idiocy to be a bit more coherent, he'd have to talk about cutting off business relationships that affect and involve the majority of his colleagues in government and a great number of other powerful people.. But why think of that when some silly little media-popular cheap shot against a specific AI company can be vomited out for the public and your constituents?
Oh, wait, I do know: he's an idiot.
Or the USA taking the UK's research on nuclear weapons during the Manhattan Project and cutting off access to it by British scientists.
I'm not even British and those are just two examples from the top of my head. The hypocrisy is absolutely infuriating...
You could write this sentence about the current President. And yet, he was elected. So much of the EOs and bills passed recently are utterly absurd — and while that perhaps stops some of them, it has far from stopped all of them.
Stop giving the GOP a free pass because it is "ridiculous". The Overton window of "ridiculous bills" is moving at a breakneck pace as it is. These people ought to be leaders. Hold them to account. Unelect them, ideally, though clearly my fellow Americans do not agree with me on that point.
Hawley looks like a clown and acts like a clown but he knows how to play the game. The only two things the guy talks about is how much he loves israel and hates china. With those credentials, he might be president one day.
ollama run deepseek-r1:671b
[1] https://forum.devtalk.com/t/how-to-run-deepseek-on-your-mac-...
Would be one option...
This statute falls under the 2M5 section of the sentencing guideline (for export violations); it has a base offense level of 14 (15-21 months) and is probation-eligible at that level. There are lots of accelerators for 2M5 offenses, but it's hard to see any of them applying to casual, or even commercial, users of DeepSeek. There are level-reducers that would apply.
(2M5 crimes can optionally be sentenced under 2B1.1, which is what most crimes we talk about on HN, particularly CFAA, are sentenced with; there the penantly would scale with financial damages. Again: hard to see how that would meaningfully apply here).
None of this is to suggest any federal prosecution for using DeepSeek would ever be reasonable. I don't even think Josh Hawley believes that. I think he just feels lonely and left out.
Many dictatorships have very broad laws that almost everyone inevitably breaks, but as long as you don't challenge the dictator or speak out against them in any way, you are safe.
But the moment you speak out against the government, suddenly the law gets enforced, since in the governments eyes, you have now committed the one true crime.
> The bill, which also prohibits the “transfer of research,” could create an unworkable environment for computer scientists who make their research public, and regularly read AI papers published by Chinese researchers.
> “Beyond just impacting people downloading models from China, the bill's penalties for the import to or export from China of AI technology and intellectual property could also potentially extend to anyone who publishes AI models or research papers on the open internet knowing they will be downloaded by people in China,” Bankston said. “Researchers are also threatened by the second half of the bill, which would directly outlaw American collaboration with researchers at basically any Chinese university or company—with a fine of up to 100 million dollars for any company that violates the prohibition, amongst other penalties.”
(I already spent the first Trump term trying to discern "malice or incompetence", it's a red-herring, the investment is unreasonable, I'm declaring it "both.")
That word has a meaning, and a model that isn’t politically aligned isn’t it.
That said, I’ve been using deep seek distilled to qwen, which should yield an incredibly censored model if they had been censoring the models, but instead yields a pretty balanced model that is more than willing to talk about Tiananmen and Xi Jinpings human rights failings.
That's what I get for hoping for a better world.
This goes to the heart of the debate here. Obviously forking an open-source model of Chinese origin is not “sending dollars and gigs of data” to China.
Hawley is a non-stupid person so he almost certainly gets this. The issue is that for a few US companies with >>$100B of market cap/valuation whose boards happen to be extremely plugged into Washington’s money machine, their strategic enemy isn’t China; it’s open-source models.
It's absurd, and while it was mostly reformed in the last 1990's, parts of it still linger around.
There's also plenty of good stories from it too, like how Bruce Schneier's "Applied Cryptography" was approved for export, but the exact source code that was it in, but on a floppy drive, was not. [2]
[1] https://en.wikipedia.org/wiki/Export_of_cryptography_from_th... [2] https://www.ka9q.net/export/
> Hawley introduced the legislation, titled the Decoupling America’s > Artificial Intelligence Capabilities from China Act, > on Wednesday of last year.
On Wednesday of last year you say.... hmmm what wrote this article
It mirrors what we saw in other industries that they basically had zero effective enforcement for, such as Cannabis sale and usage. It did not stop the sale or usage of Cannabis at all, it was very common for people to purchase and consume while illegal. But, it was "effective" at ensuring if anyone who wanted to explore that space or improve it essentially had to be a criminal and market forces were absent or distorted at best.
Fast forward and look at areas that have passed recreational sale and usage laws and they are flourishing in this respect. They have innovated the market, reduced cost, created a safer and better product, etc. In every measurable way it has improved the situation for everyone involved, including people who are basically uninvolved but benefit from the massive tax revenue that often gets redirected to schools and other social programs.
Banning DeepSeek and any other model ensures that when a kid comes up with something cool that using one of those models, he has to either make sure to keep it on a secret GitLab account or run `sed s/deepseek/muricahai/i *.py` before he shares it anywhere that might come back to him.
This is about instilling commercial fear, and stopping any US open source distributions to depend on it.
So crazier things have happened!
It seems bad for rights holders of many kinds of intellectual property (or else they wouldn’t be filing so many lawsuits).
It seems bad for progress in the field: for the most part our frontier vendors don’t contribute innovations back to the commons in anything like the way that DeepSeek has. This seems to mean that effort is duplicated at tremendous cost in a way that props up famine gouging markups for NVIDIA but little else.
It seems bad for the employees of those same vendors: they’ve been getting laid off left, right, and center with AI as a (dubious) justification.
It’s far from obvious that further growth will be powered by renewable energy, there is a lot of talk indicating that much of it will end up being coal brought back on line.
And even the investors are shaping up to be heavily in Japan and the UAE if this Stargate stuff is real.
It sounds like Mag7 shareholders maybe? And even that isn’t clear?
Maybe I’m missing something obvious, but extremely influential people in the space are selling this family of plan with visions of “wages crashing to zero”, which yeah no thanks.
A downloaded model that can be run on US servers is covered by the first amendment as free speech. This is not the same circumstance as the law that required TikTok divest of its platform, which was over spying concerns[0]. You can’t spy on Americans when data is kept on US servers.
0. Per the TikTok Supreme Court ruling:
https://www.supremecourt.gov/opinions/24pdf/24-656_ca7d.pdf
There is no doubt that, for more than 170 million Americans, TikTok offers a distinctive and expansive outlet for expression, means of engagement, and source of community. But Congress has determined that divestiture is necessary to address its well-supported national security concerns regarding TikTok’s data collection practices and relationship with a foreign adversary. For the foregoing reasons, we conclude that the challenged provisions do not violate petitioners’ First Amendment rights.
Even Google AI said that an LLM is not itself covered under free speech; only the content it produces, which itself is ironic since AI companies stole world copyrighted data to feed into their LLM software machines.
This is absolutely not settled law.
. . .
(P.s. I'm not a politician)
Also: Preventing the import of "foreign" knowledge / research seems completely insane to me, no offense, no matter how one stands on export. There is nothing justifying this, you're just kneecapping yourself.
[1] https://www.hawley.senate.gov/wp-content/uploads/2025/01/Haw...
Or it would if the courts abided by the principles they were supposed to.
" bill that would make it illegal to import or export artificial intelligence products to and from China"
Also it makes importing illegal, which means that you as a consumer wouldn't even be able to run it (ip or dns block).
-- Honorable ex Prime Minister Jim Hacker
Trump admires Xi and Elon loves China. The people at the very top are more China-friendly than the last administration.
Meanwhile we seem interested in cutting research and pissing off our neighbors… presumably opening opportunities for China.
I don’t understand what if anything these Senators are thinking.
These guys are China’s best friends.
Hopefully Europe will start closer bilateral dealing with China, and not hang itself to the future hermit kingdom of america.
In Italy, the purportedly nationalist Meloni has caved in on all issues during the Biden administration and is now courting Trump.
There needs to be a new generation of politicians who remember the more independent days from 1970-2000.
It is a disgrace that first the US led the EU into their pet conflict in Ukraine, (deliberately) ruins EU-Russia and EU-China relations and now Trump mocks the EU, demands Greenland and EU payments to Ukraine to the tune of $200 billion.
US have many faults and I would surely appreciate bigger independence for Europe. That's a fact.
But the rest is just made up nonsense. I wonder why your account was created only for this post, have one comment and is under two weeks old. Comrade gopnik.
Seems like there'd be pushback from Apple, MS, etc on that one.
> ... tenor processing unit...
I'm reading this and it can't possibly mean what it says.
Section 3 says:
(a) PROHIBITION ON IMPORTATION.—On and after the date that is 180 days after the date of the enactment of this Act, the importation into the United States of artificial intelligence or generative artificial intelligence technology or intellectual property developed or produced in the People’s Republic of China is prohibited.
(b) PROHIBITION ON EXPORT.—On and after the date that is 180 days after the date of the enactment of this Act, the export, reexport, or in-country transfer of artificial intelligence or generative artificial intelligence technology or intellectual property to or within the People’s Republic of China is prohibited.
And in Sec. 2 (6) "Intellectual property" is defined as work protected by copyright, property protected by patent, stuff which is trademarked, or trade secrets.
But any preprint on (e.g.) arxiv.org is copyrighted, hence "intellectual property" under this definition. So as written, this seems to prohibit the exchange of research in general with people in the PRC. The restrictions on AI are problematic enough, but this is just ridiculous.
Someone on LinkedIn posted how “dangerous” it was that AWS was going to make DeepSeek available and hosted on Bedrock because it would give China access to data. Not understanding that if it were hosted by AWS, China wouldn’t have access to the data.
This is like this bill where he doesn’t understand that if you “download” the model, you aren’t giving China access to your data.
"DeepSeek's Hidden Bias: How We Cut It by 76% Without Performance Loss" (2025) https://news.ycombinator.com/item?id=42868271
https://news.ycombinator.com/item?id=42891042
TIL about BBQ: Bias Benchmark for QA
"BBQ: A Hand-Built Bias Benchmark for Question Answering" (2021) https://arxiv.org/abs/2110.08193
That said, there has always been a serious imbalance with China that has yet to be addressed as far as I know. This isn't a US-vs-China thing. This applies to every country in the world. In most nations around the world China is able to buy land, property, businesses, etc. The reciprocal isn't true. We could say a similar thing about intellectual property. Good luck enforcing yours in China. Entire industries in China have been built on the back of, to be kind, borrowed IP.
One could have made the argument to look the other way 40 or 50 years ago, when China was an agrarian society in need of economic help. That is no longer the case, by far. Why is it that residents or companies from western countries cannot fully own property in China in the same way as the Chinese can do everywhere else in the world?
This concept of lack of reciprocity extends into such things as data and privacy rights and ownership. Everyone knows that any service based around user data (TikTok, AI, whatever) based in China creates 100% exposure of that data to government entities, without any level of transparency or accountability --particularly if you are not Chinese and likely worse if you are.
I think it is good and likely necessary to call China to task on these issues and apply (or continue to apply) pressure for them to open the doors to reasonable levels of reciprocity. The relationship, otherwise, is decidedly one-sided, and this means that nobody will ever trust them. Why would anyone send AI queries and data to servers in China? Or use any current or future code generator offerings to work on projects? That would be, at a minimum, suicidal.
So, while this bill is bonkers, maybe it launches a conversation that might, in a few years, drive reform that could open China and Chinese services like TikTok and AI to the world without fear of use, abuse and repercussions stemming from the nature of their society and government. To be sure, I think China would benefit immensely from a greater degree of openness.
(I just assume it's huge. I'm Spartacus and Spartacus doesn't really know about such things.)
We want it to be in this country and we're making it available. I'm going to help a lot through emergency declarations because we have an emergency. We have to get this stuff built. So, they have to produce a lot of electricity, and we'll make it possible for them to get that production done very easily at their own plants if they want, where they'll build at the plant, the AI plant, they'll build energy generation and that will be incredible.
But it's technology and artificial intelligence all made in the USA. Begin immediately, Stargate will be building the physical and virtual infrastructure to power the next generation of advancements in AI. And this will include the construction of colossal data centers, very, very massive structures. I was in the real estate business, these buildings, these are big beautiful buildings that are going to employ a lot of people and physical campuses in locations currently being scouted nationwide.
Practically zero chance to get through house and senate.
It wouldn't be any more stupid than this, and would probably solve a lot more of the US's problems...