FlexGuard: Continuous Risk Scoring for Strictness-Adaptive LLM Content Moderation (2026.acl-long)
Copied to clipboard
| Challenge: | Existing guardrail models for content moderation assume a fixed definition of harmfulness, but enforced strictness varies across platforms and evolves over time, resulting in brittle moderators. |
| Approach: | They propose a strictness-adaptive LLM moderation benchmark that enables controlled evaluation under multiple strictness regimes. |
| Outcome: | The proposed moderator performs better under one regime and under another, and is more robust under varying strictness. |
Similar Papers
LS-Guard: Adaptive Safety Guardrails Tailored to Individual LLMs (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing security guardrails built from static datasets ignore each model’s unique safety profile and often force trade-offs between safety and utility. |
| Approach: | They propose a framework for learning model-specific guardrails tailored to each LLM’s vulnerabilities. |
| Outcome: | The proposed framework significantly outperforms baseline guardrails on multiple real-world LLMs, achieving superior robustness, adaptability, and generalization. |
FLEX: A Benchmark for Evaluating Robustness of Fairness in Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing safety evaluations may overlook the inherent weaknesses of Large Language Models, despite their benefits. |
| Approach: | They propose a benchmark to evaluate the robustness of Large Language Models under extreme conditions. |
| Outcome: | The proposed approach evaluates the fairness of large language models under extreme conditions. |
XGUARD: A Graded Benchmark for Evaluating Safety Failures of Large Language Models on Extremist Content (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing safety evaluations rely on binary labels, overlooking the nuanced risk these outputs pose. |
| Approach: | They propose a framework to assess the severity of extremist content generated by Large Language Models (LLMs) it categorizes model responses into five danger levels (0–4) defined by degree of extremism endorsement . |
| Outcome: | The proposed framework categorizes model responses into five danger levels (0–4) defined by degree of extremist endorsement, enabling nuanced analysis of failure frequency and severity. |
Challenges and Remedies of Domain-Specific Classifiers as LLM Guardrails: Self-Harm as a Case Study (2025.naacl-industry)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have impressive capabilities in generating human-like text, but they pose significant risks in many domains and require guardrails throughout the lifecycle. |
| Approach: | They propose to use a self-harm detector to test the performance of LLM guardrails in real-world environments. |
| Outcome: | The proposed model performs poorly in open and closed domains and is almost unusable in the real world. |
Multitask-Bench: Unveiling and Mitigating Safety Gaps in LLMs Fine-tuning (2025.coling-main)
Copied to clipboard
| Challenge: | Recent advances in Large Language Models (LLMs) have led to their adoption across a wide range of tasks, ranging from code generation to machine translation and sentiment analysis. |
| Approach: | They propose to fine-tune LLMs on benign (non-harmful) data to ensure safe outputs. |
| Outcome: | The proposed model reduces attack success rates across a range of tasks without compromising its usefulness. |
AEGIS2.0: A Diverse AI Safety Dataset and Risks Taxonomy for Alignment of LLM Guardrails (2025.naacl-long)
Copied to clipboard
Shaona Ghosh, Prasoon Varshney, Makesh Narsimhan Sreedhar, Aishwarya Padmakumar, Traian Rebedea, Jibin Rajan Varghese, Christopher Parisien
| Challenge: | Existing safety-related content safety models are not well-suited for commercial use. |
| Approach: | They propose a taxonomy that can be used to categorize safety risks . it combines human annotations with a multi-LLM "jury" system to assess safety . they plan to open-source Aegis2.0 data and models to aid in safety guardrailing . |
| Outcome: | The proposed taxonomy can be used to assess the safety of human-LLM interactions . it can be trained on large, non-commercial datasets and is open-source . |
LLM Safety From Within: Detecting Harmful Content with Internal Representations (2026.acl-long)
Copied to clipboard
| Challenge: | State-of-the-art guard models rely on terminal-layer representations and overlook safety-relevant features encoded across internal layers. |
| Approach: | They propose a lightweight guard model that harnesses safety neurons from LLM internals without modifying the underlying model. |
| Outcome: | The proposed model outperforms open-source guard models across multiple benchmarks while using 250 fewer trainable parameters. |
Jailbreaking Attacks vs. Content Safety Filters: How Far Are We in the LLM Safety Arms Race? (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing studies have focused on the models, neglecting the full deployment pipeline . previous studies have underestimated the practical success of these attacks . |
| Approach: | They evaluate the effectiveness of jailbreak attacks targeting LLM safety alignment . they highlight critical gaps and call for further refinement of detection accuracy and usability . |
| Outcome: | The proposed attacks can detect at least one safety filter across the entire deployment pipeline. |
HiddenGuard: Fine-Grained Safe Generation with Specialized Representation Router (2026.acl-long)
Copied to clipboard
| Challenge: | Current alignment approaches rely on refusal alignment to avoid harmful content . large language models are often overly cautious or overlook subtle harmful content. |
| Approach: | They propose a framework for fine-grained safe generation in Large Language Models that enables real-time, token-level harmfulness detection and redaction without loss in capability. |
| Outcome: | The proposed framework achieves over 90% in F1 score for detecting and redacting harmful content while preserving overall utility and informativeness of the model’s responses. |
Safety Is Not Universal: The Selective Safety Trap in LLM Alignment (2026.findings-acl)
Copied to clipboard
Iago Alves Brito, Walcy Rios, Julia Soares Dollis, Diogo Fernandes Costa Silva, Arlindo Rodrigues Galvão Filho
| Challenge: | Existing safety evaluations of large language models aggregate harms under generic categories such as "Identity Hate" a bilingual benchmark identifies a selective safety trap, where defense rates vary by up to 42% within the same model solely based on the target group. |
| Approach: | They propose a bilingual adversarial benchmark to audit selective safety in large language models . defense rates vary by up to 42% within the same model solely based on target group . |
| Outcome: | The proposed benchmark identifies a selective safety trap in large language models . defense rates vary by up to 42% within the same model solely based on the target group. |