I haven’t seen this particular quirk outside GPT users, but Claude’s seems to be, “X is quietly doing work…” or some variation of that.

“You really hit the nail on the head, but the thing you said about X is doing quiet work as well.”

Your reasoning is doing quiet work. The context is doing quiet work. Everything is doing quiet work. We’re all a bunch of librarians out here, apparently.

Oh, no. No way. Those fucks were trained on every single pubic human word written, all of our art, all of our labor—for free. These companies have lobbied the government to dodge regulation and avoid accountability this entire time. Ain’t no way I’m gonna let these companies avoid the heartache they put on others.

Oh darn, is your IP theft machine losing you money on software licenses? Fuck off.

“It is imperative that the cylinder and the larger structure remain unharmed” always kills me. I don’t think I’ll ever forget that thread.

:3

I think it’s just because the answer doesn’t really contribute much. OP is asking about something specific (and kinda interesting, too!) and your response was “I don’t know, I don’t use that software”. It doesn’t really add to the conversation or help to answer the question.

Imagine if every question that got asked was responded to by the entire community, even if they had no answer, and every thread was just filled with “I dunno”.

No hate on you specifically, just thinking out loud a bit.

If you have Mac as Navi in your fan casting, I totally abandon mine because this is so much better than mine.

I cannot tell you how fast this would work on me (also fucking preach).

The Gang Solves Hyrule’s Political Crisis

Also, is that shaved Charlie as Navi? My casting would be Dennis as Link, Charlie as Ganondorf, and Mac as the fairy, for obvious reasons that the gang makes immediately clear, but all the other roles are the same. Frank would, of course, play all of the other roles as well.

does not casually creep your profiles

I was gonna say, that flies in the face of like everything I’ve heard about Korea, haha.

Love seeing this when it’s posted but that one about oil prices is aging like milk lately, hahaha.

I’m sorry, he was adroit a few months ago? Did I miss this?

Wait, wait, wait, you have weed in south Korea???

Brad Neely’s stuff always fucking slays me

I would dip my entire body in orange paint if I got one single fraction of the attention that pumpkins get. 😭

If you combine them, you sometimes also get a degradation kink from your partner commenting on it.

“Good girl.”

uncomfortable/excited reaction

“Pathetic. You can’t even accept a compliment?”

even more ridiculous reactions

In love with this new terminology you’ve gifted me. I’m going to abuse this when I describe my ideal partner.

Unironically, this is based as fuck

In their paper, they post keys that can be verified once the vulnerabilities are patched (so they aren’t just revealing exploitable issues to the world) but in the few that they demonstrated (ones that were quickly patched), it demonstrated a pretty sophisticated ability to find and exploit multiple vulnerabilities. The patches that you saw them mention are a direct result of Anthropic reporting those vulnerabilities.

The method they talk about is basically saying that they weren’t looking at old, patched code (which would mean that the model could have found vulnerability mentions on the web that others have pointed out) but rather current, actively used software. The vulnerabilities and exploits that the model found were novel, zero day (meaning as of yet they ‘undiscovered’ problems by the person/people being attacked).

I’m not a researcher though, so someone can correct any information I’ve gotten wrong here, but this is definitely not solely hype. It’s not exciting stuff (unless you just look at headlines) but the vulnerabilities they discovered are like actual problems, especially if a model like this gets into the hands of bad actors.