Why Aligned AI
We're making safer and more robust artificial intelligences by developing new fundamental methods for improving machine learning. AIs that are guaranteed to operate in your interests, including in situations where it’s not easy to determine what your interests are, is the core of alignment. This means that the AI can be trusted to perform more tasks without causing errors or needing human oversight.
Power Through Safety
Our Approach
Safer AI is more adoptable AI. If you can’t trust an AI to do what you intend for it to do, then it is not applicable for important or high risk situations. That's why we're putting safety at the heart of everything we do.
The root of the problem
Our Science
Getting AI to do what we want it to do has been the main challenge of the field since 1956. Despite recent progress, this remains unsolved. We’re getting closer with concept extrapolation.
Making AIs safer
Our Applications
We’re making AIs safer today by building tools that can be integrated into current systems. But we’re also building the safest and most aligned AI system from the ground up.




