| Sep 24, 2026 | Three papers accepted to NeurIPS 2026, on emergence during training, feature absorption in sparse autoencoders, and shift-invariant attention. |
| Aug 17, 2026 | Completed the BlueDot Impact Technical AI Safety Project, a five-week research sprint with facilitated cohort discussions. |
| Aug 10, 2026 | Attending the Iliad Intensive in Berkeley, a three-week residential research intensive on theoretical AI alignment. |
| Jun 04, 2026 | Two papers accepted to the ACL 2026 Workshop on StereACuLT, on cross-lingual bias and stereotype editing. |
| May 25, 2026 | Two papers accepted to ICML 2026 Workshops on DEMO and TAIGR: CLASP and PolicyLLM. |
| May 05, 2026 | Our position paper Multi-Agent Explainability Needs Contracts Before Methods has been accepted to ICML 2026. |