MetaBackdoor shows that LLMs can be backdoored using positional triggers like sequence length, enabling stealthy activation on clean inputs to leak system prompts or trigger malicious behavior.
Code alpaca: An instruction-following llama model for code generation
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2representative citing papers
LPO-derived Secure-Anchored task vectors raise simultaneous functional-and-secure code rates by 2.1–36 pp on six coding LLMs with near-zero inference overhead.
citing papers explorer
-
MetaBackdoor: Exploiting Positional Encoding as a Backdoor Attack Surface in LLMs
MetaBackdoor shows that LLMs can be backdoored using positional triggers like sequence length, enabling stealthy activation on clean inputs to leak system prompts or trigger malicious behavior.
-
Functional and Secure Code Generation with Task Vectors
LPO-derived Secure-Anchored task vectors raise simultaneous functional-and-secure code rates by 2.1–36 pp on six coding LLMs with near-zero inference overhead.