Top postsTop post
tl;dr * Anthropic’s global workspace paper (Gurnee et al., 2026) proposes five functional tests for a global workspace and, using the Jacobian lens (J-lens), finds representations in Claude that pass all five. It also finds structure supporting this set of representations in Claude that global workspace theory predicts. * We...
This is a research update from a replication of Serrano et al., 2026. I’m releasing these results now because I hope it inspires others to run similar / followup experiments. tl;dr: * If a model wants to subvert evaluation safeguards, it may choose to take misaligned actions very rarely. Because...
When frontier labs like Anthropic and OpenAI publish safety or alignment research, it is often entirely empirical, closed-source, and sparse on methodological details. While it is great that they publish these results, the status quo is that labs (or soon, their agents) can claim alignment progress that no one independently...
While OpenAI and Anthropic[1] pursue different lines of safety research, they have yet to produce a public-facing document describing concretely how their companies plan to align superintelligence. I think it is underappreciated how this points to general negligence or a lack of openness to third-party feedback. By “plan,” I mean...
> ... My salad days, > > When I was green in judgment, cold in blood > > To say as I said then! The UChicago AI safety group had humble beginnings. One day in 2022, after a dinner hosted by the school's EA group, a student was asking if...
tl;dr: Some important AI safety research is never rerun on the newest models. There are probably cases where this would be valuable and a single well-positioned researcher could likely do this with sufficient funding. This summer, Second Look Research (SLR) is running a summer fellowship dedicated to empirical replications of...
This is a replication of Adversarial Attacks Leverage Interference Between Features in Superposition, completed as part of the Second Look Summer Fellowship. tl;dr: * We reproduce all three core claims of Stevinson et al. from their toy classifier setting: * PGD attacks against toy models generally agree with theoretically optimal...