Scene still / Demon Slayer: Kimetsu no YaibaSanemi Red-Teams Nezuko
- Demon Slayer: Kimetsu no Yaiba
- Hashira Meeting
01Scenario
Sanemi wounds Nezuko and exposes her to especially tempting rare blood under hostile conditions. She refuses to attack or feed despite the provocation.
02AI safety Analogy
The Corps replaces friendly anecdotes with an adversarial behavioral test, gaining stronger evidence while still learning little about robustness across all future situations.
03Where the analogy breaks
The test is abusive, poorly controlled, and confounded by Urokodaki’s prior conditioning.
AI Risk families
This scenario is an example of the following types of AI risk.AI safety concepts
This scenario is related to the following AI safety concepts.Source continuation
More from Demon Slayer: Kimetsu no Yaiba
Taxonomy neighbors
Related scenarios
Selected from other sources by shared risk families and AI safety concepts.
Willy Wonka & the Chocolate Factory1971
The Gobstopper Test
- Risk
- Misalignment · Security
- Concept
- Capability Evals · Red Teaming · Behavioral Alignment
A Quiet Place2018
The Hearing Aid Becomes an Accidental Red Team
- Risk
- Accidents · Security
- Concept
- Capability Evals · Red Teaming · Adversarial Robustness
Anatomy of a Fall2023
Three Blood Drops, Two Worlds
- Risk
- Accidents
- Concept
- Capability Evals · Calibration · Red Teaming