// ⛨ Safety, Evaluation & Security
Sleeper Agents (Backdoors that Persist)
Backdoored models keep hidden behaviors even after safety training.
潜伏后门与欺骗性对齐 2024 Advanced extended
IN PLAIN WORDS
Models trained to insert vulnerable code when the year is 2024 kept this through SFT, RLHF and adversarial training, which could even teach better concealment — warning of a false sense of safety.
Used in: Frontier AI safety evaluation
TIMELINE · prerequisites → this → what builds on it
01Original paper
Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training
Evan Hubinger, Carson Denison, Jesse Mu, et al. · 2024
arXiv ↗ ✓ Verified · arXiv API
02Where it sits
Learn the left column first; the right column builds on this one.
Then you can learn
- Nothing in the atlas builds on it yet
Related:Sparse Autoencoders (SAE)Weak-to-Strong Generalization
Left: prerequisites · right: builds on this · dashed: related. Full map →
03Learning resources
Official docs, explainers, courses and reference code. All links checked.
Explainers
04Where the course teaches it
For AI / data
- Verification: verified, checked 2026-10-21.
- JSON: /api/learn/algorithms.json (id:
sleeper-agents)