A blog series explaining sheaf-theoretic reward spaces and Sheaf-Geodesic Policy Optimization, a mathematical approach to AI alignment and safety. Posts use analogies and worked examples, and link to a free paper, source code and experiment results.
Website: https://oasisx.ghost.io/
Essays and research from Princeton-affiliated authors arguing that AI is a transformative but normal technology rather than a path to superintelligence. Issues cover AI agent evaluation, the economics of automation versus augmentation, and a critical look at the AI safety movement.
Research posts from Redwood Research on catastrophic AI risk, focused on AI control, monitoring of agentic models, and red-teaming of safety systems. Readers get technical write-ups of papers, experiments and engagements with AI developers, including methods, results and open questions.
A weekly roundup of AI developments covering model releases, safety and alignment incidents, cybersecurity events and AI policy, with extended commentary on individual stories. It also includes occasional standalone deep dives, such as analysis of AI legislation and company conduct.
Transformer reports on the politics and policy of advanced AI, mixing original reporting, interviews and tests of AI models with analysis of legislation and company strategy. Readers get coverage of Congress, AI safety debates and frontier labs' behavior.
One short email a week with hand-picked newsletters from the index. Free, no spam, unsubscribe anytime.
