Backdoor attacks pose a serious threat to large language models (LLMs), but existing defenses remain fragmented, failing to provide unified defense against both dirty-label and clean-label attacks. To investigate why such fragmentation arises, we present the first systematic feature-level mechanistic analysis of LLM backdoors using sparse autoencoders (SAEs). Starting from a 2×2 comparison of clean and poisoned models on clean and triggered inputs, we trace backdoor-induced logit shifts to high-contributing SAE features and categorize them into four roles: interaction, suppressed, mixed, and weight-modified features. This taxonomy reveals systematic encoding differences: dirty-label backdoors are dominated by isolated interaction features, whereas clean-label backdoors rely more on heterogeneous mixtures of mixed and weight-modified features. These differences explain why existing defenses remain fragmented across attack paradigms. We validate this hypothesis through inference-time feature clamping, which reduces ASR to at most **10.8%** in most dirty-label settings and at most **15.4%** in the majority of clean-label settings, while preserving benign-task performance. These results show that SAE-based analysis can explain defense fragmentation and guide interpretable backdoor mitigation.

Deep research agents often rely on agreement among retrieved documents when deciding whether a claim is reliable. This heuristic breaks when the agreeing documents are not independent: several independent reports can support a claim, but several copies of the same false claim should not. We study evidence laundering, a content-only attack that publishes one false claim across paraphrased pages that appear to provide support from multiple sources. We evaluate this failure with deterministic matched corpora that present the same false claim and supporting propositions as a single source, a paraphrased correlated cluster, or an apparent independence control, with matched retrieval exposure. Experiments show that agents often assign laundered clusters credence comparable to the apparent independence control and become more willing to publish the false claim. Within the matched evaluation, a structural provenance gate substantially reduces this failure, suggesting that source independence modeling should precede answer synthesis.

While Chain-of-Thought (CoT) prompting has become a standard paradigm for eliciting complex reasoning capabilities in Large Language Models, it inadvertently exposes a new attack surface for backdoor attacks. Existing CoT backdoor attacks typically manipulate the intermediate reasoning steps to steer the model toward incorrect answers. However, these corrupted reasoning traces are readily detected by prevalent process-monitoring defenses. To address this limitation, we introduce MirageBackdoor(MirageBD), the first backdoor attack to achieve Think Well but Answer Wrong. By unlocking the model’s post-output space alongside a tailored training procedure, MirageBD enables the triggered model to preserve clean CoTs while selectively steering the final answer toward a specific target, significantly enhancing the stealthiness of the attack. Experiments show that MirageBD generally achieves over 90% attack success rate across four datasets and five models with a poison ratio of only 5%. Moreover, even under rigorous evaluations such as trigger perturbations and CoT-based detection, MirageBD maintains robust performance and stealthiness, posing a critical challenge to existing safety guardrails.

Reasoning-enhanced large language models rely on intermediate reasoning signals to solve complex, multi-step tasks, making reasoning behavior a valuable form of intellectual property. Meanwhile, knowledge distillation enables an adversary to replicate this behavior in a realistic black-box setting by repeatedly querying a deployed model on a target domain and training a local student to imitate its outputs, including reasoning traces. Existing LLM watermarks primarily operate on surface text and decoding-time token biases, and thus fail to provide reliable attribution of reasoning behavior once it is transferred through knowledge distillation. ReasMark entangles the watermark with the target-domain input distribution by selecting watermark tokens from high-frequency prompts, so distillation queries naturally activate it. It then embeds the watermark by score-conditioned losses that create a detectable reasoning-length gap for black-box verification. Comprehensive experiments across multiple LLMs, datasets, and distillation settings demonstrate that ReasMark consistently outperforms existing baselines while preserving task utility.

While Retrieval-Augmented Generation (RAG) systems are designed to enhance factual fidelity by grounding LLMs in provided sources, the application of current watermarking techniques creates a paradoxical failure mode. These methods, being inherently fact-agnostic, force the model to deviate from the very source documents it is supposed to follow. This leads to “faithfulness hallucinations"—a critical flaw where the generated output contradicts its own grounding context. Consequently, these watermarks undermine the core value of RAG, rendering even the most secure schemes untrustworthy for high-stakes applications. To resolve this RAG-specific conflict, we introduce the Dual Factual Shield (DFS) framework, a novel architecture designed to enforce knowledge loyalty. The DFS framework employs a defense-in-depth strategy through two synergistic layers: a source-anchored algorithmic safeguard that shields critical terms from the retrieved context, and prompt-based semantic guidance that protects against factual corruption. To demonstrate its effectiveness, we enhance a state-of-the-art, spoofing-aware contrastive watermarking baseline with our framework. Experiments show that our framework drastically reduces the Knowledge Corruption Rate (KCR)—a new metric we introduce—while preserving its original high security and robustness. This work establishes a new paradigm for watermarking, evolving it from merely secure to truly trustworthy. We demonstrate that traceability and truth can, and must, coexist, paving the way for the responsible deployment of traceable AI in knowledge-critical domains.