Resolution

Resolution’s strategy is to run multiple interacting research bets aimed at principled alignment rationales (for example, scalable oversight via debate-like mechanisms), and to couple those research bets with engineering and automation work intended to scale experimentation and theoretical progress. Resolution’s formation is explicitly described as bringing together researchers from the UK AISI Alignment Team and Timaeus, and the organization reports an in-person presence in the San Francisco Bay Area (Berkeley) alongside researchers working remotely in London, Melbourne, and elsewhere. Resolution also emphasizes a “whole-of-field” style integration: it expects different lines of alignment research to interact and improve the odds that at least some bets yield confidence-relevant progress. Beyond alignment theory, Resolution has also begun expanding into adjacent work such as agent foundations, explicitly continuing the spirit of the MIRI Agent Foundations tradition. Strategically, Resolution positions itself as an independent institute aimed at aligning research toward higher-confidence claims (including in contexts beyond what current empirical scaling alone can deliver) while still producing empirically grounded artifacts—such as publication outputs and, per its funding and engineering updates, large-scale “token” usage intended to support automated proof-generation or “autoformalization” workflows. Resolution’s recent public updates highlight major resourcing (a $160M grant from Coefficient Giving plus additional OpenAI API credits) and ongoing hiring across research, engineering, and operations.