tl;dr: Anthropic supports government centralized government control over models, Amazon produced a probably bogus request to pull down Mythos and Fable, so Trump pulled it down.
It’s probably bogus because no evidence of effective jailbreaks were provided, and also Fable/Mythos isn’t any more capable than OpenAI’s pre-jailbroken 5.5 offering, making it a moot point.
Anthropic can put it back up once they institute citizenship checks for their customers and ban any foreign nationals they employ from using it.
(All of the above according to Anthropic)
I’ll editorialize and say that this is blatant illegal retaliation on the part of the admin, and also that anthropic brought it on themselves with their “this model will kill us all” Mythos marketing stunt.
I guess in this story, Amazon is the useful pawn/idiot. Maybe it’ll go Shakespearean, and we’ll get some lowbrow comic relief from Bezos.
Honestly, banning SOTA LLM services is the best thing the US could do for AI.
It’d force people to run inference locally, and that’d expose the actual $/perf of the models instead of keeping it secret then propping it up with circular revenue and blatant securities fraud.
If we don’t do something like that, we won’t have much of an AI industry post-bubble.
It’s not just the US. Honda recently announced they’re not able to make internationally competitive (=== high quality and affordable) cars anymore, and abandoned their entire EV line.
I guess they’ll just milk the ICE assembly lines until they are bailed our or go under, Detroit-style.
The printing press gave us the renaissance, even though the church argued it was too dangerous to give non-clergy access to books.
Even things like universal access to guns was a net positive. It led to the end of feudalism and rise of democracy.
The sad truth is that whenever any one group of people gets a monopoly over an important technology, they use it to exploit/enslave/murder everyone they can. Look at the international news for examples from 2026.
Since the Renaissance got started before the printing press, maybe you mean the press fueled it? The idea that the church found printing dangerous seems like a conflation with events that happened during the Protestant Reformation. The Catholic Church did censor works it found heretical, including unauthorized Bible translations.
One could argue the opposite conclusion, that technology helps break monopolies, but either view depends on reductionist historical readings. The truth is somewhere in between.
Restricting things like creation of a highly infectious virus is very different from restricting books or even guns. There is no 'monopoly' over such a technology, as a use of the technology will inevitably harm the creators themselves.
Restrictions on high end biology, chemistry would leave overwhelming number of use cases of LLMs unaffected - no need to ban open weight LLMs. Such restrictions can be even more effective, if it is coupled to researchers getting early access to see the possible problems and have an opportunity to prevent the outbreak or create new vaccines well in advance.
Restrictions are not enabling monopolies. The opposite is true, if a LLM engineered virus or other harmful technology is let loose, public opinion can very quickly swing towards draconian regulation. (see nuclear power after Chernobyl).
I'm amazed we didn't have the same moral panic when the web became popular. billions of people suddenly had access to knowledge about how to create dangerous viruses! sites like Wikipedia don't even check that you're a US citizen before letting you access pages on recombinant DNA and genetic engineering! the articles on sarin and VX nerve gas include syntheses!
Wikipedia is a presentation of partial selection of biology textbooks and research papers, not using them as a collective brain to generate new artifacts.
There is a big difference between having a large bookshelf of programming language/networking/OS manuals and the ability to generate a functional software product which previously required a hundred or more developers. Even a hundred developers may not be able to find a subtle exploit in code which requires a tedious scan of millions of lines. Computer security hacks can be much less of a problem in comparison to exploits in biology.
Also, even Wikipedia (and public resources in general) have restrictions - there is information dangerous enough to be not published. In the 1930's itself, Szilard (who discovered the chain reaction) and Bohr advocated for restrictions on openly publishing research on uranium fission.
Speaking practically your hypothetical is a scenario that requires somebody that is proactively interested in, and theoretically capable of, making a e.g. dangerous virus, yet are unwilling/unable to do so without a chatbot. How many people might this possibly apply to? I think the number is literally zero.
I also don't entirely understand your comment, because your latter parts do not follow from your lead. You're 100% right that somebody who's not extremely capable messing with this stuff is overwhelmingly likely to just hurt themselves. And somebody relying on a chatbot to guide them in dealing with this sort of tech? Yeah, they're gonna win a Darwin Award.
---
I also think there's an entirely different, yet also compelling argument, against censorship. Local LLMs already exist and are advancing rapidly. There will come a time, probably in the relatively near future, when the state of the art big system and a decent uncensored local system will become practically indistinguishable in terms of capability. So not only will people be able to do this locally, but you lose something big in the process.
The reality is that our interactions with LLMs are 100% being actively surveilled, regardless of privacy promises of the companies involved. At the minimum, every chat is making it's way over to the NSA's Utah data center, one way or the other. Some guy trying to do something significantly malicious using an LLM is little more than a gift to the authorities, but this is only true with centralized/online uncensored services. Push people onto local models to do nefarious stuff, and law enforcement is blinding themselves.
>Speaking practically your hypothetical is a scenario that requires somebody that is proactively interested in, and theoretically capable of, making a e.g. dangerous virus, yet are unwilling/unable to do so without a chatbot. How many people might this possibly apply to? I think the number is literally zero.
I don't disagree with the rest of your post, but this doesn't seem correct.
I think I'd phrase it that there probably already exist, or will exist, people with the inclination to cause global mass death, but don't have the knowledge or ability to manufacture a virus to achieve this.
If it just a mundane chatbot, the discussion is moot. But, we already have AI making breakthroughs in research and approaching the abilities do science just like a scientist does. (The last two paragraphs of your comment also assume such a high capability scenario).
Imagine giving the access, to whoever wants it, to a scientist who may not have many fresh insights, but has the advantage of a huge memory containing all the scientific literature in their mind, the standard patterns of deductions, and the ability to work at a very fast pace 24/7. They could identify vulnerabilities in biological mechanisms, just like AI identifies security flaws in code today.
---
Regarding hurting themselves, I was not referring to someone who is too dumb to follow lab safety precautions, but someone who has a nihilistic mindset. State actors and militia use weapons to take over and enjoy the power they acquire - they dont want to get killed by a deadly virus(unless they engineer and selectively apply the vaccine before they release the weapon - but this is very hard to keep secret). Someone who is nihilistic wont have such reservations on using the weapon even if it destroys them eventually.
Regarding restrictions on API LLMs leading to use of local LLMs, it is the local LLMs which will be used anyway (once they have the capability). That we live in a mass surveillance envirnoment is common knowledge. The bottleneck, where restrictions can be applied, is not inference but training which requires hundreds of millions of dollars. Chinese scientists have themselves spoken about AI safety concerns and it is indeed a threat to China just like anyone else.
Also, restricting high end weapons ability does not interfere with 99.9% of LLM usage (open-weights or proprietary) - so it need not interfere with business strategy.
I like the premise but it sounds like something where the overhead in trying to track & manage that would be overly burdensome for all parties until you just forced more reasonable terms on when material enters the public domain in the first place, at which point such a system wouldn't really be needed anyways. The last thing I want to see to try to clean up public access to work is even more complex rules and systems being layered on top of the existing system.
Until they mandate open firmware, it’s basically a moot point. The most severe problems modern cars have are software problems with the ADAS systems, where the car increasingly becomes erratic with age. (Slamming on brakes when there’s a tailgater, swerving into ditches, over double yellows, etc).
In other news, try obtaining OEM brakes for a twenty year old mustang. They making them.
Unless I’m missing something the “early access” benefit is just plain evil.
They’ll expand to new cities, but not allow residents of those cities to use the service. Instead, you need to be from SF or some other place they operate, and pay a monthly fee.
- Arnold bought a fleet of mobile hospitals that would have been perfect for covid response, but the next governor didn’t want to pay 1% the fleet cost per year to maintain it, so he scrapped it.
- Under Obama, SARS v1 was stopped by US health workers that Trump fired because it was a “bad deal”. In the absence of that team, we got SARS v2, which was renamed to COVID 19.
There’s also the related category of “never blamed for fixing problems poorly, creating even bigger problems”.
Thanks to 9/11, plane cockpits can now be locked from the inside. Now, we have examples of commercial passenger airline pilots locking the doors and committing mass-murder-suicide by plane crash.
For some reason, these stories don’t make the news.
Background reading:
https://www.anthropic.com/news/fable-mythos-access
tl;dr: Anthropic supports government centralized government control over models, Amazon produced a probably bogus request to pull down Mythos and Fable, so Trump pulled it down.
It’s probably bogus because no evidence of effective jailbreaks were provided, and also Fable/Mythos isn’t any more capable than OpenAI’s pre-jailbroken 5.5 offering, making it a moot point.
Anthropic can put it back up once they institute citizenship checks for their customers and ban any foreign nationals they employ from using it.
(All of the above according to Anthropic)
I’ll editorialize and say that this is blatant illegal retaliation on the part of the admin, and also that anthropic brought it on themselves with their “this model will kill us all” Mythos marketing stunt.
I guess in this story, Amazon is the useful pawn/idiot. Maybe it’ll go Shakespearean, and we’ll get some lowbrow comic relief from Bezos.
reply