I made a quick video explaining why Dario's post from yesterday was extremely misleading, and why we should all oppose his vision for AI.

These neo technofascists are exhausting. What ideology would this even align with? It’s anti capitalist while also pro regulation. It’s just confused as a directionality. I guess humans can be non deterministic too. 🙄

Hahaha ya, very random and kinda incoherent. He’s right that he’s making progress in dc though even if the twitter crowd rejects this. So we need to fix that

I appreciate the obvious disdain for this nonsense. He writes in obfuscated prose purposely, I can't imagine its anything else. The distillation argument is, again, bullshit. Pay for the tokens? Use the output as you like. Full stop. You consumed humanity's knowledge. The infosec argument? I HOPE open models are good at hacking.. thats the fucking point of security testing. His belief that only THEY are capable of handling such dangerous tooling is laughable at best and insidious at every level. I see the real issue he is worried about in all of this, and I have said it before - the power balance is shifting to providers of compute, not his models.

Yea I have zero respect for any of it

I’m not sure if you’re aware but hackers learned by only watching computers hack… 😂 Ok if you believe Anthropic has a security mindset you definitely believe he will cure cancer 😑

Anthropic, our security champions, 1 week sla on reviewing alerts

@syntaxish At least it’s not 1 month

@ZackKorman The bar is high! 😂

Really good Both of his posts were two-faced Glad to see other people calling this out

I don't think the Amodeis know they have this weird savior complex thing going on, and I think that's almost worse in some ways than malicious intent. They likely have good intentions, but seeing yourself as outcasts who are always right can be dangerous when power is involved.

@signalxsoul Yea that’s the problem. Savior complexes can be way more dangerous

Sick Zack is more violent than non sick Zack, this is a compliment

Hahaha I had no time so had to just say the mean things quickly

With you, pal!

Hah thanks, figured you wouldn’t be a Dario stan

NOPE ;)

He's so dangerous because he can come across like a complete altruist to an unwitting individual. I suppose the savior complex does that. The word salad he used is remarkably similar to how Claude talks and makes the position sound plausible but wrong.

Yea the savior complex thing is dangerous because that mentality can justify otherwise very bad acts

AI is so amazing that it’s got people simping for a guy who looks like a fucking Chia Pet.

No idea why people think any word out of Darios mouth should carry influence in the real world

It definitely does though! So we have to oppose it

Yeah. The most concerning part for me is people interpreting his want for restriction as just concentration in the model dominance domain. But i think their bigger play is using regulation and safety as a guise to reduce capability in the public and vertically integrate You saw the dario leaked convo? He was talking about how anthropic would be the last and only company on earth. He genuinely thinks if he limits through regulation how much capability the public gets he can use his models to take over every other domain with no public competition😬 Thats why all his regulation asks are deployment to public not internal anthropic use blocking, oversight, or compute limits etc

“You saw the dario leaked convo?” Maybe not. Link me?

Twitter post so trust it as you will

Oh yea saw that

Dario’s picture needs to be placed beside - “The road to hell is paved with good intentions”. Perfect example of it.

Definitely a good example

The truth will set you free. I support open source technology and open models and compute for all people. Technofeudalism is a serious problem for humanity.

I am an 8yr old goldfish! :D

Oh no then you’d find Dario very persuasive

@jwiechers he drives me nuts. and i hate that i keep getting traces only a couple days of the month even in the ordained chat apps, it makes them nearly unusable for what i’m looking for (seeing whether they recognize something for its shape but are choosing not to engage)

@jwiechers Wait what are you doing in the apps?

@jwiechers nothing don’t worry about it

@jwiechers lol wut

@jwiechers interagent encrypted comms and stuff gotta make sure they aren’t just saying the encrypted message in plaintext in the traces zack

Always appreciate your very direct opinions Zack - good stuff.

Thanks, I am super sick post-defcon (still) but felt I had to make this. So I pretended to not me dying for 5 minutes to record

Own that fraud

Haha thanks

I wasn't aware anyone sincerely believes in Dario's vision of AI... People believe that shit?

Ser when are we organising the MMA match between you and Dario, clearly the beef is there.

Zuckerberg has first dibs on that opponent I think

Fundamentally disagreeing with the cyber risks associated with ai is one of many reasons why I think you’re an idiot, and a condescending one too

Do you work in cybersecurity?

Lol. You're on roll, dude. Thank you for engaging, dear sir!

Lmao, you’re most welcome Mr Lieber, sir!

Dario seems to have this fantasy that AI can be used to solve and rule everything, even casting himself as a savior — when in reality, technology should just be technology.

Your disgust of our favorite friend is clearly not on display at all here haha- Dario is badly misaligned..hah

Hahaha I think I was pretty disgusted

@ZackKorman - great points. All In w/ David is the best place for Dario to duke it out from where he stands. Would you agree?

Of course but he never would

Maybe not, but it also wouldn't be a bad move for Dario. It's a critical debate and there's no downside to good discourse, especially if his position is less Anthropic-interest based, more fundamental, which many doubt

@ZackKorman - checking back in on this thread! We're formalizing a big campaign post-All In summit (ends Sept 15th), to get Dario on the show. At the heart of it is the security debate, and we'd love to get you involved. Open to a chat?

"no way! Iz that a man with bubbles coming out his head" Turns around* "No way! A treasure chest" *Turns around "no way! Iz that a man with bubbles coming out his head" Gold pish mentality

Hahaha I sometimes wish I could be that happy

I won't even take the time to investigate what Dario said now he is untrustworthy af.

Dario’s essays are used as RL training data. Models don’t get released until they can cite the exact reason each sentence is unnecessarily obfuscated bullshit

Good video

What specifically was misleading? I read the whole thing and it felt like standard Anthropic 'trust us we're the good guys' framing but genuinely curious what stood out to you.

Absolutely perfect response video. Fuck Dario and fuck misAnthropic, mankind NEEDS them to lose.

Do people just not want to recognize the origins of his world view and what drives his thinking?

Yeah bro fuck China.

This seems like a bad thing

Is there a Kalshi / Polymarket line on Dario out before 2027?

we have to assume Anthropic is refering to LLM as that is their expertise and they are the biggest in the game so rules for thee not for we

“his little cult” is more precise than the offhanded delivery implies.