Si No Se Puede
Linkpost note: I originally avoided cross-posting this to Lesswrong because I felt readers would likely…
Category: Opinion
Claude knows who you are
Kelsey Piper noticed that Opus 4.7 is the first model which can identify her from…
Five approaches to evaluating training-based control measures
Training-based control studies how effective different training methods are at constraining the behavior of misaligned…
Idea Economics
It happened again… I have an idea for a project. It’s a super cool project.…
What Are with All the Different GPT-5 Variants in Codex and How Are They Actually Different? (or are they even?)
OpenAI has been consistently making it hard to understand what each of their different models…
Morality without Consciousness
A recent post by @Linch, “The Fourth World,” gets at an important implication of consciousness—…
AI for decision advice
We’ve written about why we think AI character — the behaviour of AI systems —…
Variations On Tree Reconstruction
Credit - Minna SundbergMethodology unearths and explores regularities, structural properties of the world that enable our methods…
3 years of being on birth control
Warning: This is completely based on my own experiences; please do your own research before…
Prompted CoT Early Exit Undermines the Monitoring Benefits of CoT Uncontrollability
Code: github.com/ElleNajt/controllability tldr: Yueh-Han et al. (2026) showed that models have a harder time making…