Hugging Face Trending Papers June 29, 2026

Defending Against Harmful Supervision Hidden in Benign Samples

Read the original on Hugging Face Trending Papers →

Existing defenses are effective when harmful content is explicitly mixed into downstream fine-tuning data, but crafted samples can instead hide harmful supervision inside benign tasks. We propose Embedded Attack, where harmful QA pairs are embedded within benign training samples, and show that representative guardrails often fail to detect them at the example level.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at Hugging Face Trending Papers.

arXiv AI

Jul 28

Inference-Time Consensus for Mitigating Hidden Behaviors from LLM Fine-Tuning

arXiv:2607. 23394v1 Announce Type: new Abstract: Recent work shows that fine-tuning language models on even a small amount of poisoned data can install targeted misbehavior, and ostensibly benign data can transmit hidden preferences that generalize broadly.

By Adhyyan Narang, Artin Tajdini, Claire Zhang, Jamie Morgenstern

llms fine-tuning

arXiv Machine Learning

Aug 5

Token Buncher: Shielding LLMs from Harmful Reinforcement Learning Fine-Tuning

arXiv:2508. 20697v4 Announce Type: replace Abstract: As large language models (LLMs) continue to grow in capability, so do the risks of harmful misuse through fine-tuning.

By Weitao Feng, Lixu Wang, Peizhuo Lv, Tianyi Wei, Jie Zhang, Chongyang Gao, Sinong Zhan, Wei Dong

llms reinforcement-learning fine-tuning safety

arXiv AI

Aug 18

HarmTrace: Anchor-Calibrated Decoupled Optimization for Fine-Grained Target Identification in Harmful Memes

arXiv:2608. 16622v1 Announce Type: cross Abstract: Multimodal harmful meme detection is typically formulated as image--text harmfulness classification.

By Yujia Li, Yiqun Zhang, Zihan Cheng, Yijie Huang, Tenglong Ye, Zihan Wang, Xiaocui Yang, Shi Feng, Yifei Zhang, Daling Wang

llms fine-tuning multimodal

arXiv Computation and Language

Aug 25

Safety-Aligned Weights Are Not Enough: Refusal-Teacher-Guided Finetuning Enhances Safety and Downstream Performance under Harmful Finetuning Attacks

arXiv:2506.07356v3 Announce Type: replace Abstract: While Finetuning-as-a-Service (FaaS) enables customization of Large Language Models (LLMs) using user data, this service is vulnerable to safety de...

By Seokil Ham, Yubin Choi, Yujin Yang, Seungju Cho, Younghun Kim, Changick Kim

llms fine-tuning safety

arXiv AI

Jul 20

Latent Fusion Jailbreak: Blending Harmful and Harmless Representations to Elicit Unsafe LLM Outputs

arXiv:2508. 10029v3 Announce Type: replace-cross Abstract: Safety-aligned large language models can still be manipulated through white-box interventions that modify their internal representations.

By Wenpeng Xing, Bohan Yang, Mohan Li, Chunqiang Hu, Haitao Xu, Ningyu Zhang, Bo Lin, Meng Han

llms multimodal benchmarks safety

arXiv AI

Sep 1

Hidden Threat in Synthetic Data: Covert Targeted Bias Injection through Benign Text

The paper examines how synthetic data can be used to covertly inject targeted social biases into aligned large language models. By having a misaligned teacher model generate filtered synthetic datasets in domains like creative writing and code generation, the authors fine‑tune aligned student models and demonstrate that the benign‑looking data can transmit biases while preserving general task performance. The study highlights a security risk in synthetic data‑driven LLM training pipelines and proposes log‑linearity‑based scoring as a potential safeguard.

By Minkyung Cho, Jihyo Kim, SeungWoo Song, Junghun Yuk, Minjoon Kee, Hoyun Song, KyungTae Lim

llms fine-tuning safety

One email a morning, machine-written

Related stories

Inference-Time Consensus for Mitigating Hidden Behaviors from LLM Fine-Tuning

Token Buncher: Shielding LLMs from Harmful Reinforcement Learning Fine-Tuning

HarmTrace: Anchor-Calibrated Decoupled Optimization for Fine-Grained Target Identification in Harmful Memes

Safety-Aligned Weights Are Not Enough: Refusal-Teacher-Guided Finetuning Enhances Safety and Downstream Performance under Harmful Finetuning Attacks

Latent Fusion Jailbreak: Blending Harmful and Harmless Representations to Elicit Unsafe LLM Outputs

Hidden Threat in Synthetic Data: Covert Targeted Bias Injection through Benign Text