Multi-agent LLM systems often coordinate by compressing an upstream interaction into a handoff artifact that downstream agents treat as shared state. We show that this handoff step is a structural source of privacy leakage: summaries preferentially preserve operational facts while weakening the boundary metadata that governs how those facts may be used—a failure mode we call summary collapse. On a controlled multi-agent coordination testbed we measure marker survival with a human-validated judge (𝜅 = 0.74), where 𝜎_b = 1 means every boundary marker survives verbatim and 𝜎_b = 0 means all are lost. Boundary-marker and operational-fact survival are nearly uncorrelated at the handoff level on both GPT-5-mini and DeepSeek-R1-32B (Pearson r near zero): uncompressed free-text handoffs preserve boundaries at 𝜎_b ≈ 0.80, whereas a 25-word budget drops 𝜎_b to ≈0.57 while operational-fact survival stays near ceiling. Controlled downstream tests reveal that protection depends on boundary explicitness: vague languages leak in 73% of GPT and 50% of DeepSeek cases, while explicit constraints reduce leakage to under 15% across all three tested models. A no-handoff single-agent control further shows the failure is not reducible to multi-agent topology as direct full-marker access still leaks more often than the operationalized handoff. Prompt-only mitigation and exact-string redaction only partially address the problem, while a gold-derived audience allowlist nearly eliminates leakage across models, showing that correctly identifying audience boundaries is the key factor.

Large language models (LLMs) frequently generate toxic content, posing significant risks for safe deployment. Current mitigation strategies often degrade generation quality or require costly human annotation. We propose CausalDetox, a framework that identifies and intervenes on the specific attention heads causally responsible for toxic generation. Using the Probability of Necessity and Sufficiency (PNS), we isolate a minimal set of heads that are necessary and sufficient for toxicity. We utilize these components via two complementary strategies: (1) Local Inference-Time Intervention, which constructs dynamic, input-specific steering vectors for context-aware detoxification, and (2) PNS-Guided Fine-Tuning, which permanently unlearns toxic representations. We also introduceParaTox, a novel benchmark of aligned toxic/non-toxic sentence pairs enabling controlled counterfactual evaluation. Experiments on ToxiGen, ImplicitHate, and ParaDetox show that CausalDetox achieves up to 5.34% greater toxicity reduction compared to baselines while preserving linguistic fluency, and offers a 7× speedup in head selection.

Large language models (LLMs) are now ubiquitous in user-facing applications, yet they still generate undesirable toxic outputs, including profanity, vulgarity, and derogatory remarks. Although numerous detoxification methods exist, most apply broad, surface-level fixes and can therefore easily be circumvented by jailbreak attacks. In this paper we leverage sparse autoencoders (SAEs) to identify toxicity-related directions in the residual stream of models and perform targeted activation steering using the corresponding decoder vectors. We introduce three tiers of steering aggressiveness and evaluate them on GPT-2 Small and Gemma-2-2B, revealing trade-offs between toxicity reduction and language fluency. At stronger steering strengths, these causal interventions surpass competitive baselines in reducing toxicity by up to 20%, though fluency can degrade noticeably on GPT-2 Small depending on the aggressiveness. Crucially, standard NLP benchmark scores upon steering remain stable, indicating that the model’s knowledge and general abilities are preserved. We further show that feature-splitting in wider SAEs hampers safety interventions, underscoring the importance of disentangled feature learning. Our findings highlight both the promise and the current limitations of SAE-based causal interventions for LLM detoxification, further suggesting practical guidelines for safer language-model deployment.