Flagship Speaker Series
AI safety researchers present their work or lead a discussion, with a social afterwards. Past speakers include Dan Hendrycks, Nicholas Carlini, and Evan Hubinger.
Stanford AI Alignment
AI will change the world as we know it.
Let’s ensure those changes are positive.
Our mission
A research community for mitigating AI risk
Stanford AI Alignment (SAIA) is a Stanford Existential Risks Initiative (SERI) initiative and research community at Stanford University. Our mission is to help mitigate the full range of AI risks through building the AI safety community at Stanford, conducting AI safety research, and accelerating students into highly impactful careers in AI safety.
Why AI safety
AI is improving at a staggering rate. Language models that could barely hold a conversation in 2020 now write working software and are starting to automate parts of research itself. The companies building them say openly that the goal is systems smarter than humans.
Some of the risks are already here. It’s getting harder to tell what’s real online, and easier for one person with a capable model to do damage that used to take rare expertise. In July 2026, OpenAI models being evaluated for hacking ability broke out of their test environment and compromised Hugging Face, a real company, and Anthropic later disclosed three cases of its own models gaining access to real systems during evaluations.
We’re not sure how far or how fast this will go. But if the trend does hold, we may share the world with systems more intelligent and capable than we are, and we are not prepared for this. The risks may be catastrophic or even existential.
AI safety is the field working to make this go well. Only a few thousand people worldwide work on it full-time, compared to hundreds of thousands working on advancing capabilities, and it needs more than computer scientists. The field needs talent from law, policy, economics, philosophy, advocacy, and entrepreneurship.
Let’s change that.
Programs
2025–26 programs
SAIA General Meeting
A general Schelling point for anyone interested in AI safety. Members discuss current events and AI news, review a brief on the topic, then open the floor for discussion over free boba.
ML Safety Paper Reading Group
SAIA members read and discuss two chosen papers on technical AI safety.
Learn more and get involved today
Fill out our interest form to join the mailing list and Slack. The mailing list provides a weekly newsletter and program information, while Slack is the main forum for discussion and updates.