arXiv subjects

Mona T. Diab

Publications and source records attributed to Mona T. Diab.

11 recordsLinked to original sources

arXiv2026

Pick Your Poison: Learning to Select Poison Sets for Stronger LLM Backdoor Attacks

Backdoor poisoning attacks add poisoned examples to otherwise-clean finetuning data, pairing a trigger with a target behavior that the model learns to produce when the trigger appears. Existing evaluations typically fix the number of poisoned examples and sample them at random from a candidate pool. We show that this can severely underestimate worst-case vulnerability: across three LLaMA-3-8B backdoor settings, holding the model, clean data, and poison count fixed, attack success ranges from 3% to 80% depending only on which poison set is chosen. We formalize poison selection as oracle-budgeted set optimization and introduce SAILS (Set-level Audit-Informed Iterative Learned Selection), which learns a set scorer from a few hundred finetune-and-evaluate runs, ranks millions of candidate sets, and audits only a small shortlist. SAILS improves held-out attack success by 30 percentage points on average over the strongest influence baselines, transfers from small-scale to full-scale finetuning, and extends to code-generation, agentic, and API-only backdoors.

cs.LG↗

arXiv2026

Decoy Direction Optimization: A Post-Hoc Defense Against LLM Abliteration

Safety guardrails in open-weight language models can be readily bypassed using Refusal Feature Ablation (RFA), a technique that identifies and projects out a linear refusal direction from the residual stream, often achieving a high attack success rate (ASR) while preserving model capability. Defending against these attacks typically requires computationally expensive safety finetuning for every new checkpoint. We introduce Decoy Direction Optimization (DDO), a fast, post-hoc weight-editing defense that requires no base-model finetuning. Our approach is based on a simple mechanistic insight: ablation attacks rely on contrastive estimators to find the refusal direction. Rather than trying to hide the true refusal circuitry, DDO actively injects a high-magnitude, nonlinear decoy signal into the network's MLP neurons. When an attacker attempts to locate the refusal direction, the decoy corrupts their estimator, tricking them into ablating a harmless orthogonal feature while the actual safety mechanism remains intact. We prove a spectral bound formalizing this effect and evaluate DDO across six model families, achieving <10% ASR under standard RFA. On Llama-3-8B-Instruct, DDO remains comparable to trained defenses under adaptive multi-phase attacks (65% vs. 58% worst-case ASR) and reduces Heretic weight-level attack ASR from 88.7% to 18%, all at 30 to 450 times lower optimization cost per configuration than the trained baselines.

cs.LG↗

arXiv2026

DSPA: Dynamic SAE Steering for Data-Efficient Preference Alignment

Preference alignment is usually achieved by weight-updating training on preference data, which adds substantial alignment-stage compute and provides limited mechanistic visibility. We propose Dynamic SAE Steering for Preference Alignment (DSPA), an inference-time method that makes sparse autoencoder (SAE) steering prompt-conditional. From preference triples, DSPA computes a conditional-difference map linking prompt features to generation-control features; during decoding, it modifies only token-active latents, without base-model weight updates. Across Gemma-2-2B/9B and Qwen3-8B, DSPA improves MT-Bench and is competitive on AlpacaEval while preserving multiple-choice accuracy. Under restricted preference data, DSPA remains robust and can rival the two-stage RAHF-SCIT pipeline while requiring up to $4.47\times$ fewer alignment-stage FLOPs. Finally, we audit the SAE features DSPA modifies, finding that preference directions are dominated by discourse and stylistic signals, and provide theory clarifying the conditional-difference map estimate and when top-$k$ ablation is principled.

cs.LG↗

arXiv2026

CCBENCH: Assessing LLM Cultural Competence via Implicitly Signaled Norms using Health Queries

To interact with users fairly and without stereotyping, AI models must display cultural competency, i.e., the ability to infer and adapt to a user's implicitly signaled cultural values, rather than relying on static demographic traits. We introduce CCBENCH, a framework for evaluating cultural competency in large language models (LLMs), treating culture as a continuum of norm adherence states rather than as a binary state of cultural belongingness. As a case study on health, we create CCBENCH-Health, which includes 60 theoretically grounded personas exhibiting varied norm-adherence states across six cultures, each engaging in 18 realistic dialogues. Each persona is evaluated on 52 authentic healthcare questions drawn from real user forums, yielding 3,120 unique interactions. Benchmarking five leading models reveals that even the best achieve culturally appropriate responses only 20-30% of the time. When explicitly prompted to focus on culturally relevant cues from the conversational history (CoT), performance improves modestly by 3-5% on average. We find that models perform best when personas avoid cultural norms rather than follow them, revealing a persistent asymmetry, suggesting a preference in the models to align with built-in biases than adapt to cultural cues. This is especially observed in the Afghan context (Avg: 8.8%), where cultural cues rarely yield appropriate health advice. Finally, we find that models sometimes adapt more readily to implicit, cultural conversational styles than to explicitly stated cultural practices, though this varies across cultures.

cs.CY↗

arXiv2026

On the Persistent Effects of Lexicality in Large Language Models

Representations extracted from large language models (LLMs) play an important role in many downstream applications. However, the structure of these representations is often influenced by lexical overlap rather than semantic content. Our understanding of the relationship between this lexical influence and semantic content, and its implications for downstream tasks, remains limited. In this work, we investigate representations to quantify the effect of lexical overlap relative to semantic content. We consider several adversarial semantic stress tests and further connect our findings to the information theory perspective. We find that lexical influence extends across the depth of models, consistently across architectures, training regimes, and objective functions, including the models trained for semantic similarity. Moreover, we observe a mid-depth region in which both lexical and semantic signals degrade simultaneously, indicating a transitional regime where representations are poor for both surface form and meaning. We further demonstrate the effect of lexical influence on downstream uses of LLMs using summarization and model editing as a case study.

cs.CL↗

arXiv2026

Pando: Do Interpretability Methods Work When Models Won't Explain Themselves?

Mechanistic interpretability is often motivated for alignment auditing, where a model's verbal explanations can be absent, incomplete, or misleading. Yet many evaluations do not control whether black-box prompting alone can recover the target behavior, so apparent gains from white-box tools may reflect elicitation rather than internal signal; we call this the elicitation confounder. We introduce Pando, a model-organism benchmark that breaks this confound via an explanation axis: models are trained to produce either faithful explanations of the true rule, no explanation, or confident but unfaithful explanations of a disjoint distractor rule. Across 720 finetuned models implementing hidden decision-tree rules, agents predict held-out model decisions from $10$ labeled query-response pairs, optionally augmented with one interpretability tool output. When explanations are faithful, black-box elicitation matches or exceeds all white-box methods; when explanations are absent or misleading, gradient-based attribution improves accuracy by 3-5 percentage points, and relevance patching, RelP, gives the largest gains, while logit lens, sparse autoencoders, and circuit tracing provide no reliable benefit. Variance decomposition suggests gradients track decision computation, which fields causally drive the output, whereas other readouts are dominated by task representation, biases toward field identity and value. We release all models, code, and evaluation infrastructure.

cs.LG↗

arXiv2025

RefusalBench: Generative Evaluation of Selective Refusal in Grounded Language Models

The ability of language models in RAG systems to selectively refuse to answer based on flawed context is critical for safety, yet remains a significant failure point. Our large-scale study reveals that even frontier models struggle in this setting, with refusal accuracy dropping below 50% on multi-document tasks, while exhibiting either dangerous overconfidence or overcaution. Static benchmarks fail to reliably evaluate this capability, as models exploit dataset-specific artifacts and memorize test instances. We introduce RefusalBench, a generative methodology that programmatically creates diagnostic test cases through controlled linguistic perturbation. Our framework employs 176 distinct perturbation strategies across six categories of informational uncertainty and three intensity levels. Evaluation of over 30 models uncovers systematic failure patterns: refusal comprises separable detection and categorization skills, and neither scale nor extended reasoning improves performance. We find that selective refusal is a trainable, alignment-sensitive capability, offering a clear path for improvement. We release two benchmarks -- RefusalBench-NQ (single document) and RefusalBench-GaRAGe (multi-document) -- and our complete generation framework to enable continued, dynamic evaluation of this critical capability.

cs.CL↗

arXiv2025

EVALUESTEER: Measuring Reward Model Steerability Towards Values and Preferences

As large language models (LLMs) are deployed globally, creating pluralistic systems that can accommodate the diverse preferences and values of users worldwide becomes essential. We introduce EVALUESTEER, a benchmark to measure LLMs' and reward models' (RMs) steerability towards users' value and stylistic preference profiles grounded in psychology and human-LLM interaction literature. To address the gap in existing datasets that do not support controlled evaluations of RM steering, we synthetically generated 165,888 preference pairs -- systematically varying pairs along 4 value dimensions (traditional, secular-rational, survival, and self-expression) and 4 style dimensions (verbosity, readability, confidence, and warmth). We use EVALUESTEER to evaluate whether, given a user profile and a pair of candidate value-laden and style-laden responses, LLMs and RMs are able to select the output that aligns with the user's preferences. We evaluate six open-source and proprietary LLMs and RMs under eleven systematic prompting conditions and six preference comparison scenarios. Notably, our results show that, when given the user's full profile of values and stylistic preferences, the best models achieve <75% accuracy at choosing the correct response, in contrast to >99% accuracy when only relevant style and value preferences are provided. EVALUESTEER thus highlights the limitations of current RMs at identifying and adapting to relevant user profile information, and provides a challenging testbed for developing RMs that can be steered towards diverse human values and preferences.

cs.CL↗

arXiv2025

Humanizing Machines: Rethinking LLM Anthropomorphism Through a Multi-Level Framework of Design

Large Language Models (LLMs) increasingly exhibit \textbf{anthropomorphism} characteristics -- human-like qualities portrayed across their outlook, language, behavior, and reasoning functions. Such characteristics enable more intuitive and engaging human-AI interactions. However, current research on anthropomorphism remains predominantly risk-focused, emphasizing over-trust and user deception while offering limited design guidance. We argue that anthropomorphism should instead be treated as a \emph{concept of design} that can be intentionally tuned to support user goals. Drawing from multiple disciplines, we propose that the anthropomorphism of an LLM-based artifact should reflect the interaction between artifact designers and interpreters. This interaction is facilitated by cues embedded in the artifact by the designers and the (cognitive) responses of the interpreters to the cues. Cues are categorized into four dimensions: \textit{perceptive, linguistic, behavioral}, and \textit{cognitive}. By analyzing the manifestation and effectiveness of each cue, we provide a unified taxonomy with actionable levers for practitioners. Consequently, we advocate for function-oriented evaluations of anthropomorphic design.

cs.CL↗

arXiv2025

Position: Mechanistic Interpretability Should Prioritize Feature Consistency in SAEs

Sparse Autoencoders (SAEs) are a prominent tool in mechanistic interpretability (MI) for decomposing neural network activations into interpretable features. However, the aspiration to identify a canonical set of features is challenged by the observed inconsistency of learned SAE features across different training runs, undermining the reliability and efficiency of MI research. This position paper argues that mechanistic interpretability should prioritize feature consistency in SAEs -- the reliable convergence to equivalent feature sets across independent runs. We propose using the Pairwise Dictionary Mean Correlation Coefficient (PW-MCC) as a practical metric to operationalize consistency and demonstrate that high levels are achievable (0.80 for TopK SAEs on LLM activations) with appropriate architectural choices. Our contributions include detailing the benefits of prioritizing consistency; providing theoretical grounding and synthetic validation using a model organism, which verifies PW-MCC as a reliable proxy for ground-truth recovery; and extending these findings to real-world LLM data, where high feature consistency strongly correlates with the semantic similarity of learned feature explanations. We call for a community-wide shift towards systematically measuring feature consistency to foster robust cumulative progress in MI.

cs.LG↗

arXiv2022

Towards Responsible Natural Language Annotation for the Varieties of Arabic

When building NLP models, there is a tendency to aim for broader coverage, often overlooking cultural and (socio)linguistic nuance. In this position paper, we make the case for care and attention to such nuances, particularly in dataset annotation, as well as the inclusion of cultural and linguistic expertise in the process. We present a playbook for responsible dataset creation for polyglossic, multidialectal languages. This work is informed by a study on Arabic annotation of social media content.

cs.CL↗