AI is moving fast. We don't know how it'll play out, and we don't think anyone really does. The readings, courses, and talks below are the ones that actually shaped how we think about this. Start wherever something looks interesting and follow the threads.
These are the risks researchers in the field actually spend their time on. Not killer robots — specific failure modes with proposed solutions and active work behind them. If you want a sense of why people take this seriously, this is where to start.
We can build AI systems that perform impressively on benchmarks. We can't reliably write down what we actually want them to do, and we can't reliably check whether they're doing it. The worry is that we deploy something powerful with goals that drift from what we meant, and by the time we notice, the system has enough influence to resist being shut down or retrained.
Last updated: October 22, 2025
These curricula and reading lists enable you to dive deeper into AI safety through independent learning.
Your career command center for AI safety. Upload your CV to get started instantly, then let our AI agent build a rich profile through conversation.
Smart Notifications
Get notified when new roles, grants, or programs match your profile — so you never miss an opportunity
CV to Profile in Seconds
Upload your CV and get an instant profile — our AI extracts your skills, experience, and interests automatically
Agentic Profile Enrichment
An AI agent chats with you to uncover goals, preferences, and context that no CV captures — building a richer profile over time
