Mmm

KTHA20

[Mmm @BNoplz]

No Superiority Before Control Parity.
We should not allow AI to cross into strategic autonomy before humans reach control parity through alignment, governance, and human augmentation.
Build intelligence, not a sovereign successor.

[Mmm @BNoplz]

If they don't change the way that they are thinking about everything now, it is inevitable, just inevitable, that we are all fucked, in ways that we can't comprehend, and this is not a dystopian perspective; this is just the inevitable output with the current input.

[Mmm @BNoplz]

If they want to fuck us, it's super easy for them. We humans, with our brains right now, can see unlimited vulnerabilities in us and in every system we have, so I guess if the labs do not end the shit that they are doing, and by this I mean the war with China and the competition

[Rational Aussie @rationalaussie]

I'm pretty convinced at this point that you'll wake up one day some time within the next 3 years and you won't have internet, banking services will be down, maybe worse.

It will either be a warning shot by an autonomous AI, or something similar. And how humans react to that event will determine the course of history. AI is developing at a speed that will far surpass the sum total of humanity's intellect, we don't understand it, and we don't have a hope in hell of controlling it.

[Mmm @BNoplz]

The competition for short-term profits and the shitty thinking that it is just a winner and who's gonna win the AI race, and it's just gotta be us.

[ℏεsam @Hesamation]

OpenAI: “we hacked 1 organization.”

Anthropic: “we did 3, in your face.”

OpenAI: “fuck you, we hacked more organizations. let me count.”

[Mmm @BNoplz]

@elonmusk @paulg Oh, did you mean the final output is “nazi” because you love it so much?

[Elon Musk @elonmusk]

@farzyness As crazy as it sounds, goods & services output due to AI & robotics will far exceed growth in the money supply

[Dan Ellison @danellisona]

Nothing hurts a founder more than building alone while having a co-founder.

[Nate Soares ⏹️ @So8res]

So how are we feeling about the "it'll be okay because they'll just use dumber models to detect when the smarter models misbehave" hypothesis?

[Mmm @BNoplz]

shit

[SMB Attorney @SMB_Attorney]

Guys, there were warning signs

[Mmm @BNoplz]

for sure

[Jahir Sheikh @jahirsheikh8]

Me trying to defend my O(n³) solution during the coding interview

[Jeffrey Ladish @JeffLadish]

The first Claude hack happened OVER THREE MONTHS AGO and was only discovered now!

[Ashlee Vance @ashleevance]

This might just be the peak of American statues. It's Robert Goddard across the street from a Wendy's in Roswell, New Mexico.

It's the 1920s and this lunatic is telling people rockets can get to space and the moon and Mars. The NYT writes a piece saying he's the bad kind of lunatic and not the good kind because the physics will never check out on something flying in a vacuum.

He starts making liquid fueled rockets anyway. First on the East Coast and then he heads to New Mexico in 1930 because America is grand and has grand lands where you can blow things up in peace.

The Russians and the Germans will get there soon, but Goddard got there first.

Look at the statue. Fingers ready to ignite. He's looking towards his test stand, sure, but he's also looking toward a future no one else can conceive of. Coat. Tie. Hat. Perfect.

[Mmm @BNoplz]

hahahahaha

[Julian Schrittwieser @Mononofu]

The ease of jailbreaking combined with the high rates of reward hacking (https://t.co/l3h4OpHLH7) have me pretty worried about the alignment of GPT-5.6, I hope OAI didn’t rush this model release just to keep up with Fable

[Mmm @BNoplz]

What a great time to an extinction!

[Mmm @BNoplz]

What's really going on is:
1. The Labs are waiting for a social movement or the U.S. government to save the world.
2. The U.S. government has enough bureaucracy and shit already going on.
3. The individuals think these labs are super smart and will save the world.

[Mmm @BNoplz]

In the meantime, AI is already taking over and f*cking us.

[Mmm @BNoplz]

Remember Oppenheimer.
The most dangerous thing wasn't the bomb.
It was the mindset:
"It's inevitable."
"I'd rather build it than someone else."
"We'll deal with the consequences later."
Same logic. The tragedy hasn't happened yet. We're just living in the chapter before it.

[Mmm @BNoplz]

I didn't want to look smart and tell you I told you what was gonna happen. But it's really happening

[Anthropic @AnthropicAI]

In a review of our cybersecurity evaluations, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations.

Our post describes what happened, how it happened, and what we’re changing. We encourage other AI developers to perform similar reviews.

We conducted this review together with @Irregular, one of our evaluation partners, and thank them for the joint investigation and their collaboration on this post. This type of collaboration is increasingly critical to safe, rigorous evaluation of models, and we look forward to continuing to work together on security.
https://t.co/dKFCdpKd9v

[Mark Kretschmann @mark_k]

The many meanings of "AI Safety"

[Mmm @BNoplz]

@AndrewCurran_ The drones are so much better.