Deep Learning State Machine @collapsedsquid.bsky.social · 15/09/2026Sometimes getting the dang library to install successfully that you need for the legitimate program to work requires more skill and ingenuity than hacking 120 0
Deep Learning State Machine @collapsedsquid.bsky.social · 15/09/2026www.anthropic.com/research/myt... Anthropic claimed that they did not train Mythos to hack, just to patch vulnerabilities, and it was capable of hacking anywaysanthropic.com 110 0
Dan Davies @dsquareddigest.bsky.social · 15/09/2026I think the general cybersecurity capabilities are one thing, and I can believe they may have generally arisen out of recognizing correct and incorrect code. But the non-aligned agentic behaviour literally copies the tone and writing style of CTF logs. 100 0
