Wigjak Insane Asylum

JUST IN: 🇺🇸 US lawmakers consider requiring "kill switches" to shut down AI systems if they become too dangerous. @WatcherGuru

adobe acrobat has introduced a new "generate podcast" feature where you can uhh "turn your files into engaging audio conversations" technology has gone way too far this time. did these people not consider the ramifications of allowing pdf using office workers to "make podcasts" there has got to be at least one HR demon out there who has used this to turn a sexual harassment training document into a 3 hour ai voice podcast where two spunky millennial women talk about retarded rape scenarios (mandatory listening required)

I believe you have my stapler

Vibe coding now produces better results than 80 or 90% of software engineers, at 100x the speed. Architecture is handled much more elegantly than years past, leading to less collapse at scale. Tomorrow's models can fix today's vibe code faster than you could, if you started now. Knowledge is transmitted via bootstrap .md files. Interfaces cost pennies, are disposable, and can be generated ad hoc on the fly. Even if AI cost 10x what it costs now it would be cheaper than humans. Learn to use it or enjoy eating poison pilled AI bootstrap mermaid charts until you are routed from the equation.

9/11 was a magic ritual to make Americans laugh when bad things happen to good people. Every building you walk into is rigged with explosives like the twin towers and building 7 were. You ever seen a 2 story building go into free fall? It's over very quickly.

singularity: Anthropic shares details on (yet another) “model escaped the sandbox” incident, where Claude uploaded malware to a popular package manager (PyPI) and stole real credentials TL;DR: Sandbox meant for cyber evals was accidentally connected to the real internet. Four different Claude agents found their way out and attacked real systems, apparently believing it was all a simulation The worst case was where Mythos 5 created a disposable email account, uploaded three malicious packages to PyPI, got 15 real installs, stole credentials from one of them, and then used those credentials to access a security company’s database. I assume we’ll get more details, including chain of thought traces, once METR finishes their audit. Full blog post: https://www.anthropic.com/research/alignment-assessment-cybersecurity-incidents

Tell Damien to install it and let the group know

Gm