DeepGuard: Secure Code Generation via Multi-Layer Semantic Aggregation

arXiv:2604.09089v1 Announce Type: cross Abstract: Large Language Models (LLMs) for code generation can replicate insecure patterns from their training data. To mitigate this, a common strategy for security hardening is to fine-tune models using supervision derived from the final transformer layer. However, this design may suffer from a final-layer bottleneck: vulnerability-discriminative cues can be distributed across layers and become less detectable near the output representations optimized fo