Skip to content
Preprint

Automatic Hard Example Synthesis with Multi-Level Agentic Data Curation

Jul 2026 · 0 citations · 52 references
Computer Science

Abstract

Multimodal Large Language Models (MLLMs) are increasingly deployed for nuanced content safety and moderation tasks, yet they remain vulnerable to adversarial attacks and out-of-distribution edge cases. Traditional active learning and manual annotation fail to scale against the complexity and volume of novel multimodal threats. In this paper, we propose an automated, agentic red-teaming framework that systematically synthesizes difficult examples using an iterative strategy that proposes novel hypotheses as well as mutating on past attempts. Leveraging a multi-agent architecture that consists of a high-reasoning Architect agent, an advanced image generator, and a multi-level verification committee of LLM raters, our system autonomously uncovers boundary-pushing violations and ambiguous policy edge cases without any human intervention. By employing these carefully synthesized adversarial examples as in-context demonstrations via test-time Retrieval, we substantially improve the target model's robustness, reducing the False Negative Rate (FNR) from 41.2% to 24.5% in a public image safety benchmark without relying on any human labeling.

View source

Similar papers

Conference Open access 2026

Powering Verifiable Learning via Automated Evolutionary Data Synthesis

This work introduces an evolutionary, task-agnostic, strategy-guided, executably-checkable data synthesis framework that, from minimal seed supervision, jointly synthesizes problems, diverse candidate solutions, and verification artifacts, and iteratively discovers strategies via a consistency-based evaluator that enforces agreement be-tween human-annotated and strategy-induced checks.

He Du, Bowen Li, Aijun Yang et al. · 0 citations
Preprint Aug 2026

Learning What to Fail On: Failure-Mode Contextual Bandits for Adversarial Data Curation

A failure-aware adversarial retrieval-augmented framework for improving robustness in natural language understanding that combines retrieval, automated validation, contextual-bandit failure selection, and controlled adversarial retraining, enables scalable robustness improvement without additional human annotation.

Roie Kazoom, Ofir Cohen, Rami Puzis et al. · 0 citations
Preprint Jul 2026

POPS: Recovering Unlearned Multi-Modality Knowledge in MLLMs with Prompt-Optimized Parameter Shaking

This paper proposes a novel adversarial strategy, namely Prompt-Optimized Parameter Shaking (POPS), aiming to recover the supposedly unlearned multi-modality knowledge from the MLLMs, exposing fundamental vulnerabilities that challenge the foundational robustness of representative MMU-based privacy protections.

Zhangheng Li, Jianing Zhu, Junyuan Hong et al. · 0 citations
Preprint Jul 2026

Position: It's Time to Optimize LLMs for Self-Consistency

This position paper observes that a wide variety of techniques designed to improve specific aspects of LM behavior-targeting properties as diverse as adversarial robustness and factual coherence-can be understood as special cases of a common "consistency optimization" procedure and addressed with a standard set of optimization tools.

Itamar Hagay Pres, Belinda Z. Li, L. Ruis et al. · 1 citation
Preprint Aug 2026

ADE: Agentic Data Evolution Framework for Human-Centered Objectives

Agentic Data Evolution is proposed, a data-centric framework that organizes synthetic supervision as evolving data snapshots through a closed-loop Observation-Variation-Selection procedure, where a steady-state admission mechanism acts as a quality ratchet that conservatively gates updates for sustained cross-round improvement.

Yang Yu, Yilin Jiang, Zexuan Fei et al. · 0 citations