Tootfinder

"""
[…] and I opened the case (politely examined by the guards of the Palace at four stages of my approach to the Red Hall) that held my ansible and some pictures. The pictures—films, photos, paintings, actives, and some cubes—were a little gallery of Man: people of Hain, Chiffewar, and the Cetians, of S and Terra and Alterra, of the Utter-mosts, Kapteyn, Ollul, Four-Taurus, Rokanan, Ensbo, Cime, Gde and Sheashel Haven … The king glanced at a couple without interest. “What’s this?”
“A person from Cime, a female.” I had to use the word that Gethenians would apply only to a person in the culminant phase of kemmer [a cyclical period of heightened sexual activity during which secondary gender characteristics emerge], the alternative being their word for a female animal.
“Permanently?”
“Yes.”
He dropped the cube and stood swinging from foot to foot, staring at me or a little past me, the firelight shifting on his face. “They’re all like that—like you?”
This was the hurdle I could not lower for them. They must, in the end, learn to take it in their stride.
“Yes. Gethenian sexual physiology, so far as we yet know, is unique among human beings.”
“So all of them, out on these other planets, are in permanent kemmer? A society of perverts? So Lord Tibe put it; I thought he was joking. Well, it may be the fact, but it’s a disgusting idea, Mr. Ai, and I don’t see why human beings here on earth should want or tolerate any dealings with creatures so monstrously different. […]”
"""
(Ursula K. Le Guin, The Left Hand of Darkness)

Hey folks, can you help me out with a little something: trying to come up with a good number of files limit for a file scan (for Kitten so you don’t hit heap memory limits if you run it from the wrong directory by mistake) and I’d love some statistics from real JavaScript/TypeScript Node.js projects of the number of files in them (excluding .git and node_modules).
Would you be kind enough to run the following command in Terminal in your largest projects and let me know what number your…

Just finished re-reading volume 1 of Yokohama Kaidashi Kikou by Hitoshi Ashinano in English official translation, checked out from one of my local libraries. It's really cool to see lots of the excellent Japanese stuff that I read through fan translations mostly online in the 2000s and 2010s not only come out as an official translation, but make their way onto library shelves for random people to discover without having to be deep into Japanese culture fandom.
YKK is of course a pretty legendary manga, and even though I'd read it before, it had been a while and the manga is so much more feelings-based than plot-based that it was at least as enjoyable the second time through. Ashinano's sweeping curves that capture the wide open landscapes of Minato are beautiful, and her characters also capture a particular feeling of freedom and kindness that's lovely to immerse oneself in. It's the kind of world where you understand immediately that the gun is just a memento and doesn't represent any kind of violence, foreshadowed or otherwise. Calm down, Chekhov, have a coffee, and get with the extremely relaxing times.
#AmReading #ReadingNow #Bookstodon

I just watched the #BlackHat #OpenAI #HuggingFace talk (#OpenAI understands it themselves. However, emergence is still no indication of actual intelligence, and the talk is again ripe with anthropomorphism.
3. Yes, it is surprising and somewhat terrifying how good these LLM agent loops have become in finding and exploiting vulnerabilities. We are seeing a much stronger version of the earlier next-generation fuzzing leap in tool capabilities.
4. The comparison with fuzzing and other tools is important for another reason: #OpenAI conducted their experiments on the live Internet. No, forcing internal test runs to go through a proxy server is *not the same as running in an isolated environment*. If there is a connection to a live production system, it is by definition not an effective sandbox. What happened to people letting their cool new fuzzers run live on non-consenting production targets on the Internet back then? They were criminally liable. Why should #OpenAI not be held liable for any damages caused by their systems?
Just because you don't understand the complexity that you built doesn't mean you aren't responsible for what it does!
5. The agent loops first successfully started exploited their proxy server on May 26th, discovered July 4th-6th as active exploitation and takeover. Credentials were revoked and the proxy server patched. "At this point, AI training and evaluations resume."
WHAT THE ACTUAL F...? You discovered emergent behavior in your overly complex system that you didn't understand, saw active exploitation of your internal systems to an extent that they were owned with admin privileges, and your response was to patch the uppermost surface holes, and then continue letting the out-of-control complex system run and continue on the same course? You didn't think to, let's see, maybe stop or at least pause the highly irresponsible experiment while figuring out - in detail - what went wrong, why, and how to avoid the same thing happening again on a systematic level?
6. On July 8th, completely unsurprisingly, the LLM agent loops continue to do the same thing and find another surface hole that hasn't been patched yet to take over again. Why should this have stopped? You haven't done any root cause analysis on the system level. Why do you expect that the problem should have stopped?
7. It takes another 11 days to discover that this is happening again. So you turned the system that had broken something back on again without a detailed root cause analysis and then didn't even watch carefully? I can't even...
8. And no, the response is not to fight fire with fire. Complexity on the attacker side (#OpenAI is the attacker, not the defender - they are the guilty perpetrator, not the innocent victim of circumstance) should be fought with *reduced* attack surface on the defender systems. Adding LLM agent loops that the "frontier" companies themselves quite obviously have no control over to already brittle systems with the hope of auto-patching your way out of vulnerabilities does not seem like a wise course of action. You don't mitigate complexity with even more complexity. The next 2 years will be ... exciting - and your best bet is going to be to disable all dependencies and complex interactions that your production systems don't absolutely require.