R.I.P.
๐ป
Ghosted
TwoHamsters: Benchmarking Multi-Concept Compositional Unsafety in Text-to-Image Models
April 17, 2026 ยท Grace Period ยท + Add venue
Authors
Chaoshuo Zhang, Yibo Liang, Mengke Tian, Chenhao Lin, Zhengyu Zhao, Le Yang, Chong Zhang, Yang Zhang, Chao Shen
arXiv ID
2604.15967
Category
cs.CR: Cryptography & Security
Cross-listed
cs.CV
Citations
0
Abstract
Despite the remarkable synthesis capabilities of text-to-image (T2I) models, safeguarding them against content violations remains a persistent challenge. Existing safety alignments primarily focus on explicit malicious concepts, often overlooking the subtle yet critical risks of compositional semantics. To address this oversight, we identify and formalize a novel vulnerability: Multi-Concept Compositional Unsafety (MCCU), where unsafe semantics stem from the implicit associations of individually benign concepts. Based on this formulation, we introduce TwoHamsters, a comprehensive benchmark comprising 17.5k prompts curated to probe MCCU vulnerabilities. Through a rigorous evaluation of 10 state-of-the-art models and 16 defense mechanisms, our analysis yields 8 pivotal insights. In particular, we demonstrate that current T2I models and defense mechanisms face severe MCCU risks: on TwoHamsters, FLUX achieves an MCCU generation success rate of 99.52%, while LLaVA-Guard only attains a recall of 41.06%, highlighting a critical limitation of the current paradigm for managing hazardous compositional generation.
Community Contributions
Found the code? Know the venue? Think something is wrong? Let us know!
๐ Similar Papers
In the same crypt โ Cryptography & Security
R.I.P.
๐ป
Ghosted
The Limitations of Deep Learning in Adversarial Settings
R.I.P.
๐ป
Ghosted
Distillation as a Defense to Adversarial Perturbations against Deep Neural Networks
R.I.P.
๐ป
Ghosted
Spectre Attacks: Exploiting Speculative Execution
R.I.P.
๐ป
Ghosted
How To Backdoor Federated Learning
R.I.P.
๐ป
Ghosted