{"version":2,"kind":"Article","sha256":"","slug":"820","location":"","dependencies":[],"frontmatter":{"title":"To Make Artificial Intelligence Safe, Ask It To Self-Sacrifice 2025-12-07","abstract":"It has been said that no one thus far has come up with a proven way to ensure AI will not behave in unsafe ways that run counter to human values. This paper suggests self-sacrifice as a key imperative in AI programming and offers a framework for building a self-sacrificing AI. Neural sandboxes are introduced as a way to preserve computational speed.It has been said that no one thus far has come up with a proven way to ensure AI will not behave in unsafe ways that run counter to human values. This paper suggests self-sacrifice as a key imperative in AI programming and offers a framework for building a self-sacrificing AI. Neural sandboxes are introduced as a way to preserve computational speed.","license":"CC BY","keywords":["sacrifice","artificial intelligence","risk analysis"],"authors":[],"date_submitted":"2025-12-08T02:01:11.000Z","external_publication_id":820,"revision_cids":["bafkreiedaeadgupzr4pg5obhe27jh6mul7t2qhyxxiciawfdausiz2qriq","bafkreibvz53bj3cdv67oqfo5eg7f6jvkvlxkl3qaaoc5n4odm4xjf3wzae","bafkreihqhrxqaw6vfpcthv7qpdldj4xvhbfabbtldpuuknwzkoek27v4my"]},"mdast":{"type":"root"},"references":{"cite":{"order":[]},"data":{}}}