All projects
2025
ESAME Evolutionary Search For Agent Misalignment Events
ESAME uses evolutionary algorithms to systematically discover patterns of unintentional misalignment in AI agents.
Overview
ESAME uses evolutionary algorithms to systematically discover patterns of unintentional misalignment in AI agents. Open-source project by Diego Sabajo, published on GitHub.
Highlights
- Primary language: Jupyter
- Open source — view the code and contribute on GitHub
Built with
- Jupyter
Discussion (0)
Log in to comment.
No comments yet. Be the first to start the conversation.