Routing Ceilings Are Domain-Independent: Structural Prior Injection in Code Security Vulnerability Detection

• URL: http://arxiv.org/abs/2607.14628v1
• Date: Thu, 16 Jul 2026 06:45:08 GMT
• Title: Routing Ceilings Are Domain-Independent: Structural Prior Injection in Code Security Vulnerability Detection
• Abstract summary: Large language models (LLMs) exhibit a well-documented gap between latent capability and consistent activation.<n>Prior work reports that structural priors raise in-distribution performance dramatically, yet collapse below the zero-shot baseline out-of-distribution.<n>We test whether this phenomenon is cross-domain by reproducing the SAIR design in source-code security vulnerability detection.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models (LLMs) exhibit a well-documented gap between latent capability and consistent activation: the router hypothesis posits that models possess the knowledge to solve a task but lack reliable internal routing to activate it. Prior work in formal mathematical reasoning (SAIR, Cázares 2026) reports that structural priors (cheatsheets) raise in-distribution performance dramatically, yet collapse below the zero-shot baseline out-of-distribution (OOD) -- and that iterative recalibration amplifies rather than corrects the collapse. We test whether this phenomenon is cross-domain by reproducing the SAIR design in source-code security vulnerability detection, evaluating three LLMs (GPT-OSS-120B, Llama-3.3-70B, Gemma-4-31B) across three vulnerability categories (CWE-798, CWE-284, and the non-CWE N+1 anti-pattern) spanning syntactic, contextual, and semantic complexity, then transferring cheatsheet-augmented prompts to real-world CVE data from VUDENC (CWE-89, CWE-22). Our findings replicate and extend SAIR: (F1) structural priors lift semantic-vulnerability recall from 20.0% to 100.0% across all models; (F2) zero-shot performance degrades along a semantic complexity gradient; (F3) the same cheatsheets that saturate synthetic performance amplify distribution-shift collapse on real CVE data (CWE-89: 100% synthetic F1 to 48.9% on VUDENC, -51.1pp); (F5) iterative recalibration produces a v2 cheatsheet that performs worse than v1 on real data, mirroring SAIR's AN45c-vs-AN38 finding. These results provide evidence that the cross-distribution trade-off surface documented in SAIR generalises to code security, and that the router hypothesis is cross-domain. We argue the structural nature of the collapse motivates distribution-aware training over prompt calibration. Code and evaluation scripts: https://github.com/bytepro-ai/bitcoder-v2-research

Related papers

• Architecture-Dependent Causal Transfer of Activation States Across Large Language Models [51.56484100374058]
  Internal language activation states can be transferred causally between different large language models.<n>Injecting projected activations into a target model during generation produces a statistically significant, preregistered causal effect.<n>We conclude that end-to-end activation-state transfer between LLMs, as currently implemented, is architecture-dependent rather than universal.
  arXiv  Detail & Related papers  (2026-08-17T09:53:27Z)
• Does Fixing Break Security? An Empirical Study of Security Degradation in Iterative LLM-Driven Infrastructure-as-Code Repair [2.4063592468412276]
  Iterative IaC repair does introduce security regressions, but the conservative, defensible rate is about 3.3% of scenarios.<n>Our findings motivate security-aware feedback-loop design and actionable iteration-budget guidance.
  arXiv  Detail & Related papers  (2026-08-13T16:01:32Z)
• From CVE to CWE: Syscall-Based HIDS Generalisation [41.99844472131922]
  In operational settings, defenders need to recognise new exploits of an already known type of weakness.<n>We empirically examine whether a one-class anomaly detector trained on the normal behaviour of a set of CVEs that share a Common Weaknession (CWE) class generalises to a different, unseen CVE inside the same class.
  arXiv  Detail & Related papers  (2026-06-21T16:34:39Z)
• Do Transformers Actually Help Intrusion Detection? A Temporal Sequence Evaluation on CIC-IDS2017 [1.2934180951771597]
  We reformulate CIC-IDS 2017 as a temporal intrusion-detection task by constructing ordered flow sequences from network conversations.<n>The central finding is that padding convention, not architecture, determines the Transformer's performance.<n>We advocate leakage-free splits, explicit padding disclosure, and sequence-aware benchmarking as standard practice in future IDS research.
  arXiv  Detail & Related papers  (2026-06-09T16:57:10Z)
• Mythos and the Unverified Cage: Z3-Based Pre-Deployment Verification for Frontier-Model Sandbox Infrastructure [0.0]
  Anthropic has not publicly characterized the escape vector; some secondary accounts hypothesize a CWE-190 arithmetic vulnerability in sandbox networking code.<n>This paper presents COBALT, a Z3 SMT-based formal verification engine for identifying CWE-190/191/195 arithmetic vulnerability patterns in C/C++ infrastructure prior to deployment.<n>Proposed: a four-layer containment framework consisting of COBALT, VERDICT, DIRECTIVE-4, and SENTINEL, mapping pre-deployment verification, pre-execution constraints, output control, and runtime monitoring to the failure modes exposed by the Mythos incident.
  arXiv  Detail & Related papers  (2026-04-22T12:28:23Z)
• Security Is Relative: Training-Free Vulnerability Detection via Multi-Agent Behavioral Contract Synthesis [14.657771106188115]
  We propose Phoenix, a training-free multi-agent framework for vulnerability detection.<n>Phoenix decomposes detection into three stages: a Semantic Slicer, a Requirement Reverse Engineer, and a Contract Judge.<n>On PrimeVul Paired, Phoenix achieves F1 = 0.825 and Pair-Correct = 64.4%, surpassing RASM-Vul (F1 = 0.668) and VulTrial (F1 = 0.563)
  arXiv  Detail & Related papers  (2026-04-21T03:02:34Z)
• Synthetic Tabular Generators Fail to Preserve Behavioral Fraud Patterns: A Benchmark on Temporal, Velocity, and Multi-Account Signals [0.0]
  We formalize a taxonomy of four behavioral fraud patterns (P1-P4) covering inter-event timing, burst structure, multi-account graph motifs, and velocity-rule trigger rates.<n>We benchmark CTGAN, TVAE, GaussianCopula, and TabularARGN on IEEE-CIS Fraud Detection and the Amazon Fraud dataset.<n>The P1-P4 framework extends to any domain with entity-level sequential data, including healthcare and network security.
  arXiv  Detail & Related papers  (2026-04-13T19:36:00Z)
• AnomalyGen: Enhancing Log-Based Anomaly Detection with Code-Guided Data Augmentation [42.87177529900358]
  AnomalyGen is a novel framework that augments training data by synthesizing labeled log sequences from source code.<n> Evaluations on HDFS and Zookeeper across 12 diverse anomaly detection models show AnomalyGen consistently improves performance.
  arXiv  Detail & Related papers  (2026-04-13T07:24:43Z)
• Verify Before You Fix: Agentic Execution Grounding for Trustworthy Cross-Language Code Analysis [0.0]
  We build a unified cross-language vulnerability lifecycle framework for software vulnerability analysis.<n>We achieve 89.84-92.02% intra-language detection accuracy and 74.43-80.12% zero-shot cross-language F1.<n>These results demonstrate that execution-grounded closed-loop reasoning is a principled and practically deployable mechanism for trustworthy LLM-driven agentic AI.
  arXiv  Detail & Related papers  (2026-04-12T20:22:23Z)
• Dynamic analysis enhances issue resolution [53.50448142467294]
  DAIRA (Dynamic Analysis-enhanced Issue Resolution Agent) is an automated repair framework that embeds dynamic analysis into the agent's reasoning cycle.<n> Driven by a Test Tracing-Driven methodology, DAIRA utilizes lightweight monitors to extract critical runtime data.<n>When powered by Gemini 3 Flash Preview, DAIRA establishes a new state-of-the-art (SOTA) performance, achieving a 79.4% resolution rate on the SWE-bench Verified dataset.
  arXiv  Detail & Related papers  (2026-03-23T14:48:54Z)
• ChainFuzzer: Greybox Fuzzing for Workflow-Level Multi-Tool Vulnerabilities in LLM Agents [42.37163874052913]
  ChainFuzzer is a framework for discovering and reproducing multi-tool vulnerabilities with auditable evidence.<n>We evaluate ChainFuzzer on 20 popular open-source LLM agent apps (998 tools)
  arXiv  Detail & Related papers  (2026-03-13T03:35:54Z)
• Lazy Layers to Make Fine-Tuned Diffusion Models More Traceable [70.77600345240867]
  A novel arbitrary-in-arbitrary-out (AIAO) strategy makes watermarks resilient to fine-tuning-based removal. Unlike the existing methods of designing a backdoor for the input/output space of diffusion models, in our method, we propose to embed the backdoor into the feature space of sampled subpaths. Our empirical studies on the MS-COCO, AFHQ, LSUN, CUB-200, and DreamBooth datasets confirm the robustness of AIAO.
  arXiv  Detail & Related papers  (2024-05-01T12:03:39Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.