Here, you can find the syllabus we'll use throughout the technical fellowship (Fall 2026). Each week lists the core readings we'll discuss together, plus "further readings" for participants to (optionally) explore independently if they're especially interested in a topic.
Week 0 is intended as an optional introduction to machine learning, to provide a more thorough understanding of machine learning (ML), deep learning (DL), reinforcement learning (RL), and transformers for the rest of the fellowship. The first meeting will discuss Week 1 readings, not Week 0 readings.
This week uses the OpenAI–Hugging Face incident as a concrete case study in how advanced AI systems can create serious safety and security challenges under present-day deployment conditions. The materials move from a broad account of the incident and its significance, to a detailed technical reconstruction, and finally to the question of how model development and release practices should change as systems acquire cyber-critical capabilities. Rather than treating the episode as straightforward proof of any single risk narrative, we will ask what it reveals about autonomous action, monitoring failures, human oversight, institutional accountability, and the difficulty of reconstructing an AI system's behavior after something has gone wrong. The central question is how the AI-safety community should update when previously hypothetical concerns begin appearing in real incidents—especially when the evidence remains incomplete, interpretations are contested, and decisions about development and deployment cannot wait for perfect certainty.