Channel
Robert Miles channel
Robert Miles AI Safety
The reference video library on alignment, control, and the open problems in AI safety, presented rigorously and with no technical prerequisites.
View the channelResources
Readings, videos, podcasts, books, and courses on AI safety curated by the community. Covers everything from foundations of the field to technical alignment, governance, and model evaluations.
Videos
Channel
Rational Animations
YouTube
An animation channel adapting classic essays on AI, rationality, and existential risk into a polished visual format.
View the channel
Channel
Doom Debates
Liron Shapira
Weekly debates and interviews on existential risk from AI, with researchers, skeptics, and policy-makers.
View the channel
Channel
Machine Learning Street Talk
MLST
Long-form interviews with front-line researchers (Stuart Russell, Yoshua Bengio, Max Tegmark) on the limits and risks of deep learning.
View the channelReadings
Reading
Preventing an AI-related catastrophe
80,000 Hours
A rigorous problem profile on the risks from advanced AI and what can be done about it.
Read the profile
Reading
AGI Safety from First Principles
Richard Ngo
A foundational sequence rebuilding the case for AGI risk from scratch, without assuming any prior literature.
Read the profile
Reading
Frontier AI Regulation
Anderljung et al. · 2023
The reference proposal for regulating frontier AI models: pre-deployment evaluations, licensing, and safety obligations for developers.
Read the profile
Reading
Core Views on AI Safety
Anthropic
Anthropic's statement on why AI safety is urgent and how the lab plans to address it in practice.
Read the profilePodcasts
Podcast
AXRP
Daniel Filan
The reference podcast for technical alignment research: long-form interviews with active researchers about their papers.
Listen to the podcast
Podcast
The 80,000 Hours Podcast
80,000 Hours
Long-form conversations with AI researchers, founders, and policy-makers on how to orient a career toward impact.
Listen to the podcast
Podcast
The Cognitive Revolution
Nathan Labenz
Weekly conversations with researchers and builders from the frontier labs on how AI is evolving and what it will take to get it right.
Listen to the podcast
Podcast
Future of Life Institute Podcast
Future of Life Institute
The institutional podcast on existential risk from AI: interviews with researchers, policy-makers, and philosophers shaping the field.
Listen to the podcastBooks
Book
The Alignment Problem
Brian Christian
An accessible narrative of the core problems in aligning AI with human values, told through the researchers facing them.
View the book
Book
Human Compatible
Stuart Russell
The argument from one of the field's pioneers for redefining AI around uncertainty about human values.
View the book
Book
Superintelligence
Nick Bostrom
The book that brought the debate about AGI risk into the public sphere and shaped much of the vocabulary the field uses today.
View the book
Book
If Anyone Builds It, Everyone Dies
Yudkowsky & Soares · 2025
The most direct case yet published for why building superintelligent AI with current techniques would end badly for everyone.
View the bookCourses
Course
The Future of AI
BlueDot Impact
A free introductory course on where AI is heading and why its safety matters.
View the course
Course
AGI Safety Course
Google DeepMind
A short course from DeepMind's safety team on the risks from human-level AI and how alignment is being approached today.
View the course
Course
ARENA
Research Engineer Accelerator
A five-week intensive curriculum on interpretability, evaluations, and RL for people already comfortable with PyTorch.
View the course
Course
AI Safety, Ethics & Society
Center for AI Safety
CAIS's open textbook and course covering the catastrophic risks from AI and the ethical frameworks for addressing them.
View the course Ecosystem map
The whole field, in one place
aisafety.com is an open directory of AI safety: an interactive map of hundreds of organizations and projects, plus communities, courses, and jobs.
Open the field mapReady for the next step?
These resources go further in community. Join a discussion group and work through them with others.