Filipe Azevedo

Filipe Azevedo is a contributor at Blindsight, writing on AI misalignment, reward hacking, and interpretability.

Posts by Filipe

1 post

May 222026MisalignmentWhat happens when an AI gets very good at the wrong thing. A primer on misalignment, reward hacking, and interpretability.Filipe Azevedo9 min read

See what AI is doing in your organization.