US bill proposes jail time for people who download DeepSeek
404media.co
404media.co
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://x.com/elonmusk/status/1880975630646612254
Speculation, but it wouldn’t surprise me if Trump’s change of heart on the TikTok ban is due to Musk’s influence.
Musk needs to stay in Beijing’s favour due to the size of Tesla’s investments in China, which Beijing could destroy in a moment if Musk overly offended them. Opposing TikTok ban helps Musk stay in Beijing’s good books, supporting it would have had the opposite effect.
"Some wonder why there was such overwhelming support for us to shut down TikTok..." (https://x.com/wideofthepost/status/1787104142982283587)
I thought this was rather reductionist until TikTok was allowed back temporarily, only for pro-Plaestine slogans to be suddenly banned when it returned.
Is it the only motivation? No. Given the above and the fact that dozens of highly popular Chinese apps continue to be allowed, does it appear likely this is a leading reason? Plausible.
I swear I have heard zero people opposed to the ban who seem to even be vaguely aware of that aspect which is the entire actual issue, and instead are completely confused and think it’s about spyware or hacking. Whether or not you agree with the ban, we need to actually be discussing the same underlying issue.
I'm sorry but I have yet to see a single person who makes this point admit that it's a form of speech suppression. This is classic First Amendment precedent. Just because the speaker is someone you don't like, or its content is anti-American (or what have you), doesn't mean it's not protected speech.
The simplest case for banning TikTok is simple reciprocity. China is no longer a market that needs the level of protectionism they currently have to develop. The second best is data privacy. Just a shame they won't apply the same to US companies yet. And then an additional reason is the one that's being used currently - risk of foreign influence. You'd hate to not ban it "because it hasn't happened yet" just for it to happen in the future. China keeping out US apps seems to have worked out great for their local industry, I'm hardly convinced it wouldn't also be a good thing for US citizens. And with the foreign-country thing tossing Constitutional issues out the window, go nuts.
> Congress shall make no law respecting an establishment of religion, or prohibiting the free exercise thereof; or abridging the freedom of speech, or of the press; or the right of the people peaceably to assemble, and to petition the Government for a redress of grievances.
The Constitution does not consider Freedom of Speech a positive right, one that can be granted to some but not others. It is a negative right against the government. It's about as broad and as absolute as you can get. This is why the right also applies to legal entities (eg: companies), even foreign legal entities. It is not limited to citizens, it is not limited to persons.
The idea that a foreign, adversarial government would have a Constitutional right to propagandize our citizenry at-scale is obviously not consistent with the societal good (or the "common defense", for that matter), so is outside of the First Amendment's scope/intent.
If you want to make an argument that it's not happening, that's one thing. But, your assertions about TikTok having some blanket "right" here are false.
Again, foreign governments don't have a Constitutional right to propagandize our citizenry.
>Enjoy your slippery slope
There's certainly a slippery slope here—and a strawman, but they ain't mine.
At this point, why even have a negative-right Freedom of Speech if you're just going to treat it like a positive right anyway? Ridiculous.
https://www.cbsnews.com/news/supreme-court-tiktok-ban/
So, feel free to take it up with them.
This is my last reply. Thanks for the chat. Take care.
"Apple has a first amendment right to be able to list any app it wants, regardless of source" seems like the only claim you could make. The Lamont case doesn't directly apply since the government isn't an intermediary between Apple and ByteDance saying "are you SURREEEE you want to get this app from them?" And the existence of restrictions on what you can send/receive through the mail make it clear that Lamont isn't a blanket "you can't regulate messages between people" restriction.
So is "hey Apple, don't list things that meet that criteria" different than "hey UPS, don't send things that meet this other criteria"?
AFAICT the Supreme Court didn't really consider it from this angle anyway and just looked at it as a regulation on corporate control, which also seems completely legitimate. Can the US gov't say "certain things require US-person-owned/controlled companies?" They do for other things already.
For someone who believes in the concept of free speech, the fact that US constitutional rights to free speech in practice almost certainly do not protect overseas business interests is really an embarrassing corner case. Even if you do genuinely consider national and foreign actors to be different kinds of entity, so that it could be ethically sound to protect one and not the other, there's a simple technical reason for constantly the distinction to be artificial: Free speech protection can be gained merely by funnelling the speech through an intermediary US citizen.
This case shows that there is a first amendment right to receive.
What is being prohibited is an adversarial government having complete control over an entity that can decide which speech is delivered to which specific audience.
We can either strive for an educated populace that can identify propaganda or we use propaganda ourselves on ourselves. Propaganda is winning.
> What is being prohibited is an adversarial government having complete control over an entity that can decide which speech is delivered to which specific audience.
That seems to imply that certain voices or ideas must be excluded to prevent this control. Isn't that, at its core, a decision that some poisonous thoughts should not be allowed to reach certain people? If not, how do you see this distinction? Does freedom of speech not include freedom to hear what others may find objectionable?
Your solution is to open the floodgates, allowing all manner of military/intel-grade psychological manipulation, bots, AI, etc. to be unleashed on our population, but try to educate the entire population to become professionals at identifying and resisting these tactics?
Seems absurd on its face. It's strange to see someone trying to make that sound like the reasonable option.
Yes that's the literally the idea of TikTok/Douyin https://www.tandfonline.com/doi/abs/10.1080/17544750.2020.17...
Banning TikTok has literally nothing to do with silencing pro-Palestinian content. For fuck’s sake, Trump is trying his best to bring it back while threatening to deport students who engage in pro-Palestinian activism.
Even when TikTok themselves tried to argue that the primary reason was to prevent foreign control over a recommendation algorithm, the Supreme Court said "nope, Congress's primary motive was the data collection."
Frankly the propaganda aspect is the least troubling part of it because you're right, people can make their own choices about information sources.
But figuring out who various military members are connected to, where and how they're moving around, and especially any leaks where people take cell phones in places they shouldn't (remember that ship that installed a Starlink illegally?), they can get quite a bit of intelligence.
> One key for him is that it’s only a possible threat. Our best intelligence, including in a briefing for Congress from the Biden administration Tuesday, is that the Chinese government has not actually done the things the ban fears.
Also, from a recent All-In interview (https://www.happyscribe.com/public/all-in-with-chamath-jason...):
> I look to Jim Himes, who is the senior Democrat on the Intelligence Committee, the ranking member. He's in what's called the Gang of Eight. He has the most exquisite access to intelligence. Jim voted against the ban. And I thought, you know what? If this guy is not seeing anything on the national security level.
> [00:44:36] There was an off the record or confidential briefing to the House Intelligence Committee. You think in that meeting, there was nothing that was very meaningful that was disclosed about TikTok?
> [00:44:45] Nothing that I had seen. Is it owned by the Chinese government? Absolutely. But is there a national security risk? I have not seen that.
But he voted yes on the subsequent vote?
https://clerk.house.gov/Votes/2024145
Moreover your factoid is misleading because it omits the fact that the chair voted for the ban, along with most of the members. Of the 25 members on the committee, only 4 voted against.
> The decision by House Republicans to include TikTok as part of a larger foreign aid package, a priority for President Joe Biden with broad congressional support for Ukraine and Israel, fast-tracked the ban after an earlier version had stalled in the Senate. A standalone bill with a shorter, six-month selling deadline passed the House in March by an overwhelming bipartisan vote as both Democrats and Republicans voiced national security concerns about the app’s owner, the Chinese technology firm ByteDance Ltd. (https://www.pbs.org/newshour/politics/possible-u-s-tiktok-ba...)
Q90 - John Nicolson: It may happen elsewhere, and I can tell you what your official TikTok response was to this leak. You did not deny that these were instructions. In fact, you confirmed that these were instructions, but what you said was that the company had changed its policy in May 2019. Previously, you instructed your moderators to take down videos critical of China, specifically talking about incidents in Tiananmen Square, separatism in Tibet, all straight out of the Chinese Communist Party playbook. You confirmed that is what your moderators did, but your defence was that you had changed your policy in May 2019.
Theo Bertram: It is highly regrettable that that is what it was, but it is not our policy today, nor has it been for a long time.
https://committees.parliament.uk/writtenevidence/13247/html/
There's a sentence
What party or politicians support individual freedom in this context? How could libertarians support banning an app?
"it is unlawful for companies in the United States to provide services to distribute, maintain, or update the social media platform TikTok, unless U. S. operation of the platform is severed from Chinese control."
If they wanted to do the opposite, they could show us videos from 10/7 (there are a lot), talk to the survivors, show us historians talking about Hamas's particularly insidious brand of extremism, show videos of them hanging gay people from cranes, etc. Whatever agenda they want to push, it's clear how they could do it.
I think they probably do this with other things too, to make us hate our government. Our chaos benefits China. I think the intelligence agencies showed Congress data they collected on this.
I admit, this is a giant mound of speculation, but even if China didn't do that, it's quite obvious they could and seems likely they would. Whether or not they've exploited TikTok in that manner is certainly debatable but I don't wish to leave them the option in the future.
because they calling themselves libertarians don't make them stand for liberty!
Libertarians might actually want authoritarianism, but it's their authority!
For some stuff like this it's hard to imagine why they wouldn't be able to make it public. You need public support if you're going to ban an app that a lot of the public likes. Almost certainly IMO, no substantive evidence existed.
I couldn’t help but lol when I read that.
I don't think the closed door intelligence was about that though. I think they probably said that the persuasive power of TikTok was dangerous to the US, but imo, that's very against the First Amendment and the idea of democracy. Americans are allowed to be influenced however which way we want. We run the government, not the other way around
I think a closed door intelligence meeting is unlikely to just be them telling Congressmen scenarios they've already heard. I think they much more likely had some receipts. Some of the Senators in that meeting (from both parties) have sponsored a bill to declassify it and share it with the public, so we may find out.
Wait, what?
Are you suggesting the bill of rights only applies to US citizens? Like, does the 5th amendment not protect foreign entities from illegal search and seizure? Does the 2nd amendment not apply to foreign security companies like Securitas?
This is speculating. In other instances, the national security apparatus has provided specific evidence when it existed which makes me believe there isn’t anything more than “XYZ could happen”.
In any case, I think we can do better than make laws based on secret evidence. This is supposed to be a democracy after all. Or something.
> not speculative
So it is speculative, right? Even your comment is just speculation.
No harm in pulling down R1 right now anyway (which is what I might have done…).
> 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...
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.
https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...
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.
Are you confused by the questions/arguments in it?
> 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?
Clearly, both are impossible to implement, wouldn't you agree? The bill talks about "AI products," as per the article, and that would be what a consumer uses, not what a researcher uses. Can legislators ban the reading of papers in America? There are 1st amendment issues there. But for protection against external states messing with our population, there's a different set of rules.... 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.
You get less jail time for committing 34 felonies.
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.
>During plea negotiations with Swartz's attorneys, the prosecutors offered to recommend a sentence of six months in a low-security prison if Swartz pled guilty to 13 federal crimes. Swartz and his lead attorney rejected the deal, opting instead for a trial where prosecutors would be forced to justify their pursuit of him.
I'm not sure how you got that impression from the original exchange of:
>>[...] You get less jail time for committing 34 felonies.
>up to 20 years. Realistically some kid downloading a model would get probation [...]
Seems I was right in that inference, given other responses to the thread since.
https://en.wikipedia.org/wiki/Capitol_Records%2C_Inc._v._Tho...
Source? My impression is that in basically all of those cases, it's either because:
1. the guy is a repeat offender and/or on probation
2. the guy decided to wanted to fight to the bitter end and they threw the book at them
I'm not saying either are justified, but the implication that someone will get 20 years just because he downloaded deepseek through ollama or whatever is still false.
“It was for two or three quarter ounces of weed!” isn’t the defense you imagine.
(The graphs in the first tweet are from page 55 of the linked PDF.)
Which isn't applicable to the law being proposed. Moreover if the idea is "threaten them with the full 20 years to serve as an example", then the mandatory minimum kinda works against that? If you catch some guy and then he serves 10 years, you haven't really proven much. You're just acting in line with expectations.
1. The RIAA sent a letter telling how many songs you were distributing and offering a settlement of around $2-5 per song. Most people at this point realized that they were in fact guilty and that the RIAA had enough evidence to prove it in court and agreed to settle.
2. A small fraction ignored this or refused to settle. The RIAA then files lawsuits in some of those cases, typically over a small number of the songs that the person was distributing rather than over all the songs being distributed.
At this point most defendants would get a lawyer and be told that they will almost certainly lose and advised to settle.
3. A handful of people ignored their lawyers (or had crappy lawyers) and plowed on. Their extraordinarily bad decision making often continued during their trial. Thomas for instance lied in court and tried to destroy evidence.
This is not wise since in these suits the plaintiff is asking for statutory damages, which is a minimum of $750 per song (not per download--if you were offering 2 songs for download and they were each download 1000 times the minimum is 2 x $750, not 2000 x $750) but can go up to $30000, and it is the jury that determines the amount in that range. You really want the jury to find you sympathetic, and lying and trying to destroy evidence doesn't help with that.
4. After the inevitable victory in most cases that got far the RIAA would again offer to settle for an amount much lower than the damages awarded by the court, although higher than their original settlement offer.
I don't know how many reached this stage, but if many did most of them came to their senses and realized that appealing would probably only make it worse.
The very small number that didn't are the ones that ended up like the two cases you cited.
Nothing works like the truth, if the US could just stop being so racist it'd be that much harder to divide us. It seems like that's fundementally against our nature though.
1. https://www.npr.org/2017/10/30/560042987/russians-targeted-u...
2. https://perspectives.ushmm.org/item/german-leaflet-for-black...
That is completely nonsense. Absolutely ludicrous.
Neither Russia nor China are homogenous. Both have large Muslim minorities (and both are be pretty racist against them). Eastern Russians are not European descended and are looked down on by Western Russians. Tibetans are not Han Chinese.
Even the Asian nations that actually are pretty "homogenous" like Japan and Korea are also famously pretty racist.
They're both the remnants of large land empires, each incorporating many ethnicities, languages, religions... They are not at all homogenous.
Like, that whole thing with the Uyghurs, what did you think that was about? Or Chechnya?
https://en.wikipedia.org/wiki/Propaganda_Poster_Art_Centre
https://www.tripadvisor.com/Attraction_Review-g308272-d18871...
If you really don't trust it, copy the algorithms without forking it. That's the big allure of OSS. You should be scutinous, but you should be able to base that scrutiny on cold hard facts.
Lawfare was tried, and it failed. I hope it dies the death it deserves instead of becoming the new normal.
It’s possible to support Harris and still be reasonable about the utter failure that the lawfare campaign against Trump has been.
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.
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.
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...
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?
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.
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.
Would be one option...
git clone --mirror ... ?Full R1 version here: https://huggingface.co/unsloth/DeepSeek-R1-GGUF/tree/main/De...
Various smaller versions: https://huggingface.co/unsloth/DeepSeek-R1-GGUF/tree/main
There is also a 'BF16' version on that page with 16-bit weights, but I don't understand what it is or where it came from, given that the original model is FP8-native.
ssh user@rsync.net git lfs install
Password:
Git LFS initialized.
... so far so good ... ssh user@rsync.net git clone https://huggingface.co/deepseek-ai/DeepSeek-R1
Password:
Cloning into 'DeepSeek-R1'...
... success.Done in the spirit of the Streisand Effect but I suppose the protest is watered down by the fact that this account is in Zurich ...
Otherwise a docker one liner to run ollama should do it.
Docker is a "core util" right.
ollama run deepseek-r1:671b
That is 404GB, but there are smaller versions too.[1] https://forum.devtalk.com/t/how-to-run-deepseek-on-your-mac-...
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.
This is why stories like this are "officially" off-topic for HN ("proposed bill" is the search query you want). The actual mechanics of US politics are such that stuff like this pops up all the time, with everybody playing the televised game, as opposed to the home game like us, understanding that none of this is actually going to happen. Us talking about it was the entire ballgame for Hawley.
(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.")
> 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.”
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.
AFAICT, there's a differentiation that should be made between the model and the mobile app. TFA, and I'm assuming the bill, too, AFAICT, don't differentiate. The model seems unequivocally speech. Whether a mobile app can be banned seems like a different question; from other reporting it seems like the mobile app is laden with lovely things like keyloggers, and there, I can see "national security" might withstand scrutiny.
If the Senator were actually concerned about American's data, he'd pass some privacy laws. But the GOP is all laissez-faire capitalists these days, and they would not want the government telling American companies that that can't sell American data for money.
. . .
(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.
You do know that Trump admitted that the invasion was provoked by NATO?
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.
Note that in (a) and (b) the third part of the "or" clause is "intellectual property". It isn't qualified as (say) "artificial intelligence intellectual property".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...