Rogue internal deployments via external APIs
Published on October 15, 2025 7:34 PM GMTtl;dr: A heavily-monitored internally deployed AI with sensitive…
Category: Opinion
Minimal Prompt Induction of Self-Talk in Base LLMs
Published on October 15, 2025 1:15 AM GMTNote: This is my first LessWrong post. I’m…
Enhancing Genomic Foundation Model Robustness through Iterative Black-Box Adversarial Training
Published on October 14, 2025 8:54 PM GMTTL;DRWe test a genomic foundation model (DNABERT-2 encoder…
Humans Are Spiky (In an LLM World)
Published on October 15, 2025 8:40 AM GMTAssessments of "general" vs "spiky" capability profiles are…
Gnashing of Teeth
Published on October 15, 2025 6:11 AM GMTJacques Callot, “The Hanging”, from “The Miseries and…
Geometric Structure of Emergent Misalignment: Evidence for Multiple Independent Directions
Published on October 15, 2025 5:45 AM GMTStatus This is late draft early pre-print of…
Situational Awareness as a Prompt for LLM Parasitism
Published on October 15, 2025 1:45 AM GMTTLDR: I believe I have had a conversation…
The sum of its parts: composing AI control protocols
Published on October 15, 2025 1:11 AM GMTThis work was supported through the MARS (Mentorship…
Postrationality: An Oral History
Published on October 14, 2025 7:18 PM GMT@Gordon Seidoh Worley is going to be giving…
Why your boss isn’t worried about AI
Published on October 14, 2025 5:58 PM GMT(a note for technical folk)[1]When it comes to…