Semantic Chameleon: Corpus-Dependent Poisoning Attacks and Defenses in RAG Systems

Retrieval-Augmented Generation (RAG) systems extend large language models (LLMs) with external knowledge sources but introduce new attack surfaces through the retrieval pipeline. In particular, adversaries can poison retrieval corpora so that malicious documents are preferentially retrieved at inference time, enabling targeted manipulation of model outputs. We study gradient-guided corpus poisoning attacks against modern RAG pipelines and evaluate retrieval-layer defenses that require no modification to the underlying LLM. We implement dual-document poisoning attacks consisting of a sleeper document and a trigger document optimized using Greedy Coordinate Gradient (GCG). In a large-scale evaluation on the Security Stack Exchange corpus (67,941 documents) with 50 attack attempts, gradient-guided poisoning achieves a 38.0 percent co-retrieval rate under pure vector retrieval. We show that a simple architectural modification, hybrid retrieval combining BM25 and vector similarity, substantially mitigates this attack. Across all 50 attacks, hybrid retrieval reduces gradient-guided attack success from 38 percent to 0 percent without modifying the model or retraining the retriever. When attackers jointly optimize payloads for both sparse and dense retrieval signals, hybrid retrieval can be partially circumvented, achieving 20-44 percent success, but still significantly raises attack difficulty relative to vector-only retrieval. Evaluation across five LLM families (GPT-5.3, GPT-4o, Claude Sonnet 4.6, Llama 4, and GPT-4o-mini) shows attack success ranging from 46.7 percent to 93.3 percent. Cross-corpus evaluation on the FEVER Wikipedia dataset (25 attacks) yields 0 percent attack success across all retrieval configurations.

Can Adversarial Code Comments Fool AI Security Reviewers -- Large-Scale Empirical Study of Comment-Based Attacks and Defenses Against LLM Code Analysis

AI-assisted code review is widely used to detect vulnerabilities before production release. Prior work shows that adversarial prompt manipulation can degrade large language model (LLM) performance in code generation. We test whether similar comment-based manipulation misleads LLMs during vulnerability detection. We build a 100-sample benchmark across Python, JavaScript, and Java, each paired with eight comment variants ranging from no comments to adversarial strategies such as authority spoofing and technical deception. Eight frontier models, five commercial and three open-source, are evaluated in 9,366 trials. Adversarial comments produce small, statistically non-significant effects on detection accuracy (McNemar exact p > 0.21; all 95 percent confidence intervals include zero). This holds for commercial models with 89 to 96 percent baseline detection and open-source models with 53 to 72 percent, despite large absolute performance gaps. Unlike generation settings where comment manipulation achieves high attack success, detection performance does not meaningfully degrade. More complex adversarial strategies offer no advantage over simple manipulative comments. We test four automated defenses across 4,646 additional trials (14,012 total). Static analysis cross-referencing performs best at 96.9 percent detection and recovers 47 percent of baseline misses. Comment stripping reduces detection for weaker models by removing helpful context. Failures concentrate on inherently difficult vulnerability classes, including race conditions, timing side channels, and complex authorization logic, rather than on adversarial comments.

Retrieval Pivot Attacks in Hybrid RAG: Measuring and Mitigating Amplified Leakage from Vector Seeds to Graph Expansion

Hybrid Retrieval-Augmented Generation (RAG) pipelines combine vector similarity search with knowledge graph expansion for multi-hop reasoning. We show that this composition introduces a distinct security failure mode: a vector-retrieved "seed" chunk can pivot via entity links into sensitive graph neighborhoods, causing cross-tenant data leakage that does not occur in vector-only retrieval. We formalize this risk as Retrieval Pivot Risk (RPR) and introduce companion metrics Leakage@k, Amplification Factor, and Pivot Depth (PD) to quantify leakage magnitude and traversal structure. We present seven Retrieval Pivot Attacks that exploit the vector-to-graph boundary and show that adversarial injection is not required: naturally shared entities create cross-tenant pivot paths organically. Across a synthetic multi-tenant enterprise corpus and the Enron email corpus, the undefended hybrid pipeline exhibits high pivot risk (RPR up to 0.95) with multiple unauthorized items returned per query. Leakage consistently appears at PD=2, which we attribute to the bipartite chunk-entity topology and formalize as a proposition. We then show that enforcing authorization at a single location, the graph expansion boundary, eliminates measured leakage (RPR near 0) across both corpora, all attack variants, and label forgery rates up to 10 percent, with minimal overhead. Our results indicate the root cause is boundary enforcement, not inherently complex defenses: two individually secure retrieval components can compose into an insecure system unless authorization is re-checked at the transition point.

TRYLOCK: Defense-in-Depth Against LLM Jailbreaks via Layered Preference and Representation Engineering

Large language models remain vulnerable to jailbreak attacks, and single-layer defenses often trade security for usability. We present TRYLOCK, the first defense-in-depth architecture that combines four heterogeneous mechanisms across the inference stack: weight-level safety alignment via DPO, activation-level control via Representation Engineering (RepE) steering, adaptive steering strength selected by a lightweight sidecar classifier, and input canonicalization to neutralize encoding-based bypasses. On Mistral-7B-Instruct evaluated against a 249-prompt attack set spanning five attack families, TRYLOCK achieves 88.0% relative ASR reduction (46.5% to 5.6%), with each layer contributing unique coverage: RepE blocks 36% of attacks that bypass DPO alone, while canonicalization catches 14% of encoding attacks that evade both. We discover a non-monotonic steering phenomenon -- intermediate strength (alpha=1.0) degrades safety below baseline -- and provide mechanistic hypotheses explaining RepE-DPO interference. The adaptive sidecar reduces over-refusal from 60% to 48% while maintaining identical attack defense, demonstrating that security and usability need not be mutually exclusive. We release all components -- trained adapters, steering vectors, sidecar classifier, preference pairs, and complete evaluation methodology -- enabling full reproducibility.

SecureCode v2.0: A Production-Grade Dataset for Training Security-Aware Code Generation Models

AI assistants produce vulnerable code in 45% of security-relevant scenarios, introducing flaws into production systems at scale. Yet existing secure coding datasets fall short. They lack incident grounding, don't provide the scale modern training requires, and miss the operational security context developers need for production deployments. We present SecureCode v2.0, a production-grade dataset of 1,215 security-focused coding examples that passed structural validation and expert security review. Every example ties to actual documented security incidents with CVE references, provides vulnerable and secure implementations, demonstrates concrete attacks, and includes defense-in-depth operational guidance. The dataset covers 11 vulnerability categories (complete OWASP Top 10:2025 plus AI/ML Security Threats) across 11 languages (Python, JavaScript, Java, Go, PHP, C#, TypeScript, Ruby, Rust, Kotlin, and YAML for infrastructure-as-code). Our quality assurance framework ensures complete incident grounding. Each example includes SIEM integration strategies, infrastructure hardening recommendations (Docker, AppArmor, WAF configurations), and testing approaches using language-appropriate frameworks. The dataset uses a 4-turn conversational structure mirroring actual developer-AI interactions, escalating from basic implementations to advanced security considerations and defense-in-depth guidance. Our contributions: (1) 1,215 rigorously validated examples split into 989 training, 122 validation, and 104 test sets, (2) an automated validation framework ensuring dataset consistency, (3) a 4-turn conversational structure capturing realistic security workflows, (4) comprehensive operational security guidance with SIEM integration strategies, (5) complete language-specific implementation fidelity, and (6) open-source release of data, validation tools, and benchmarking protocols.