As artificial intelligence models advance in capabilities1, we expect them to have an increasingly profound impact on our society2. It is essential that this impact is positive, and that the decisions made by these systems are transparent, reliable, and accountable3 to the people affected by them.
We believe that reducing the risks associated with advanced AI models4 is one of the most important challenges of our time. We also believe it is an open and exciting problem5, with ample opportunities for more researchers to advance in this field6.
BAISH's mission is to support students in entering this field and conducting research on this topic.
Get InvolvedJoin us for our upcoming AI safety events, discussions, and workshops. All events are open to the community.
Intensive 1-month course on mechanistic interpretability methods.
Starts end of April 2025 Learn More →Weekly meetings to discuss recent papers, concepts, and developments in AI safety.
Every Friday @ 3pm Learn More →An 8-week guided course covering the essential concepts in AI alignment and safety.
Starting in the second semester Learn More →