Security professionals petition the US government to lift export restrictions on Fable and Mythos, arguing they lack unique capabilities

Here the letter says that pulling the best capabilities away from defenders is *dangerous*. But...

Anyway it makes no sense. There's no argument here, just a collection of random statements that contradict each other. Here's the whole thing: https://freefable.org/

Here they acknowledge that Fable didn't even work for cyber defenders. The vast majority of defenders already didn't have these "best capabilities" and that seems fine with everyone signing this?

And then they say that if the US government is going to regulate models, it is important that the regulations be decided democratically.

That's fine but kinda weird to be like "only Anthropic gets to make unilateral and arbitrary decisions on who gets access". I don't even disagree with that in principle, government should be transparent and fair in a way private companies don't have to be, but it's certainly not about security.

@ZackKorman Straight from the jump there's a huge problem when it says it should be based on scientific evaluations and the authors of this letter don't seem to realize that the CEO of Anthropic thinks his product is alive and has hired a psychiatrist for it. 😳

And here they're like "Mythos isn't even uniquely good at security stuff" so then why is removing it so dangerous?

@ZackKorman It just buys into their marketing. People need to engage their thinkin' brains I swear.

They /just/ admitted OpenAI ChatGPT 5.5 is basically as good as Fable

So just go use that. Anthropic made their bed they can lie in it.

They repeat again that this took the best models away from defenders even though they seemingly said the opposite earlier and also most defenders didn't even have access to Mythos.

@ZackKorman No please don't block Fable I need it to defend systems. My proof that it's not dangerous is that I couldn't even use it to defend systems!

@ZackKorman So you support the govt making arbitrary decisions backed by nothing but vibes? Or do you maybe like this decision because it makes your own company viable again? This one? https://embroidery.io/

And this part is just weird. Kinda shatters the idea of independence when you start talking about what Anthropic is doing and why they are doing it. Like... what?

@DecryptedTech Yea what the hell am I reading lol. It's like 8 people all made statements, they don't even fit together.

@ZackKorman You hit the nail on the head here. There is no counter-point, it is all just "we like Anthropic" ... Like an Amicus Brief, but without and new perspectives or other things that would go into one.

@akses_0x00 The letter even says that! I don’t get it!

@ZackKorman The argument is that the govt shouldn't be playing politics with the most important tech in American history. It's really fucking simple.

Respected ex-GOOG security eng @NielsProvos, reflects the sentiment well. This action is going to drive companies to move to (Chinese) open weight models for their critical systems, lest an arbitrary decision on a Friday afternoon yanks your AI SOC: https://www.provos.org/p/case-for-open-weight-models/

@ZackKorman Gib me haxing robot. I need it for science.

@Nadsec11 I tried that already they say no "to do crime" isn't a good enough reason

@HeidyKhlaaf Exactly! The letter literally argues BOTH of those points. Like what you just wrote is a summary of exactly what the letter says. Which is hilarious

@ZackKorman That's basically what it is.