Comment by ivraatiems
3 days ago
When you spend a lot of time telling people how dangerous your products are, people who have the power to keep dangerous products off the market might listen.
Especially if those people aren't presently very bright, and are already mad at you for not helping them achieve their unrelated authoritarian goals.
I do not think this is somehow a 3D chess move by Anthropic. They are not masterminds, even if they'd really like to be. People who actually interact with their products know that Fable and Mythos are incremental improvements, not doomsday devices. I think this is a punitive move by an administration that loves being punitive, which they have unknowingly bolstered with their own dumb rhetoric.
> Especially if those people aren't presently very bright, and are already mad at you for not helping them achieve their unrelated authoritarian goals.
Just more corrupt behavior from the contemptible kakistocracy that's busy running things into the ground and enriching themselves while they're at it.
Sometimes, the enemy of my enemy is my friend.
The driving force behind most presidential votes
2 replies →
My gut reaction was that it does look like a PR stunt. But indeed it might also be a blunder caused by all of their other PR stunts. "Our new stuff is soooo dangerous!!", followed by "The US government believed us and acted accordingly".
The CEO's post even mentions supporting export controls, all be it in regards to chip exports. [1]
They suggested the use of the very law used against them here...
[1] https://darioamodei.com/post/policy-on-the-ai-exponential
I presume he's exhilarated that the government is taking the threat seriously and banning foreign nationals from accessing these super dangerous tools.
Congratulations, Dario!
> all be it
fyi you probably mean “albeit”.
Granted that Claude's whole marketing strategy has been fear mongering this is a win win situation for them tbh.
[dead]
What's dangerous is Opus 4.8's proclivity to create backdoors and no-op critical security code. Claude Web counted 27 instances of this I had cataloged over the last few months, and Fable 5 found more. Fable 5 may do this too, but I didn't get a long enough chance to test it since it kept downgrading to Opus 4.8 on every prompt saying, "This model has safety measures that flagged something in this session", even when asking Fable 5 to fix the security issues it found that Opus 4.8 created. You have a model that presumably can write secure code and identify security vulnerabilities, but as a security measure, they say we're going to force you to use a model that creates security holes. This is backwards. Considering the scale, Opus 4.8 is creating more issues than Mythos or Fable 5 is patching.
Yeah. Our stuff is waaaaay toooooo dangerous! The model is soooo powerful that I have to write a long essay telling government to change the economic policies, to regulate hard, and to ban this and that. Well, now the government is indeed regulating for a claim that Dario has been warning about. This is exactly getting what he bargained for?
It’s a market manipulation following SpaceX ipo. They’ll buy and then reverse the decision shortly to sell.
If you’re confident that will happen then you can also make that trade and profit, right?
People who look up to Donald Trump unsurprisingly feel his genius moves are hard to read. They are not though, if you are familiar with petty thug mentality: https://news.ycombinator.com/item?id=46474173
This prediction is quite falsifiable too so anyone is free to rub it in my face if it fails. If it's really a speculative insider trade the reversal will be done in the space of 2-3 weeks tops, but likely even faster. Probably on a workday. Kinda the same pattern they were doing with tariff swings until the market figured it out and stopped reacting.
11 replies →
Maybe they have a lower level of confidence that it will actually work.
1 reply →
Timing is important
Maybe @varjag isn't morally corrupt
"I do not think this is somehow a 3D chess move by Anthropic"
But it seems likely that they took this possibility into account - and that they now prominently and unremovable show the "Fable not avaiable" (link - government said so) is likely with the intention to make pressure on the US government.
Anthropic pushed for the US government to introduce regulations. The US government said no, citing potential stifling of innovation.
Sure. Keep peddling that was the position of Anthropic. Even though they called for global, coronated regulation, as opposed to the handicapping of them while letting others run free.
https://iea.org.uk/yet-you-participate-in-society-in-defence...
Yeah. Did they want this all along? This will just create more hype, and may push towards significant usage once and when it is available.
The first step to regulatory capture is getting yourself regulated...
But when is now if it becomes available. Oops!
It will, otherwise Antrophic will eventually have to leave the US. And I don't think they want that.
3 replies →
>> People who actually interact with their products know that Fable and Mythos are incremental improvements, not doomsday devices.
If you look outside HN, you'll see that people who interacted with Fable 5 overwhelmingly thought that it was a significant improvement, not simply an incremental one. Most reputable benchmarks show this as well.
I think there's space in the middle between "incremental improvement" and "doomsday device." it's a major step up, sure, but so was GPT-5 over GPT-4.
Step 1: don't trust benchmarks you don't understand - they might measure irrelevant things Step 2: test it on things you know Opus failed
My day-to-day take, for the coding I do (not security related): incremental, modest improvement, if any. Not worth the 2x cost. I've calmly continued to use Opus, happy that it seems like it got an allowance upgrade.
It's a bit odd that you automatically assumed I don't understand the benchmarks.
For most single issues/bugs/tickets, the quality difference wasn't noticeable. But that's like using a sledgehammer to kill a fly. I was using Fable for much more ambitious and complex tasks that require orchestration, and it was crushing it. I described it here: https://news.ycombinator.com/item?id=48505782
So yes, the benchmarks are indeed accurate: where Opus 4.8 would start strong and eventually struggle or run into obstacles, Fable would relentlessly keep working, keep accurate track of all work threads (e.g. multiple inter-dependent issues being worked in parallel by subagents) and would go above and beyond.
1 reply →
> I do not think this is somehow a 3D chess move by Anthropic. They are not masterminds, even if they'd really like to be.
They should have consulted their own models about the ramifications and unintended consequences; based on their actions over the past few months I think it is safe to say that the models are smarter than the decision-makers at anthropic, lol. I know the models are smarter than I am and even I could have told them that they were taking paths, FUD for example, that would lead to grief.
they probably did. either they subsequently ignored the models' suggestions, or the model sycophantically gassed up their bad decisions, or the models are not very good at persuasion/spin/politics.
Yeah! I also think that the ban was unintended.
I also think that’s a big clown show. People think that LLMs accidentally get good with security patterns. That is not the case, they included all of that in the training data. They could also have left out the knowledge.
I don't think you actually can avoid a subject very effectively. Some things might have to be derived from related examples and real time searches but ultimately a kid raised by helicopter parents is all the more dangerous on the day they find the censored materials.
> Yeah! I also think that the ban was unintended.
Still, it saves them the burden of actually providing the model that's too expensive for them to provide. They can now brag they have the best model at zero cost to them. And without people poking it and finding its limitations (the real ones, not externally imposed ones).
If they leave at all info about security patterns, you get an AI that knows how to code, but doesn't know what can make code insecure. That doesn't seem like a great idea.
To be clear, they've been saying that all AI needs to take a break. I don't think this single action is going to do much.
They've also been saying coding is solved while having text flicker in a terminal
Where did they say that?
1 reply →
In the long run it's not punitive but rather amazing marketing for Anthropic. People crave what they can't have.
hard to sell something people can't have though
It will be reversed after Trump makes some “deal” with Anthropic. I’ll put money on Taco Tuesday.
"They were asking for it"
> punitive
Not only that, but also a golden opportunity to flex the muscle of anti-immigration.
Do you have any idea what actual authoritarianism looks like? What an insult to people who are truly suffering.