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.
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.
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.
arXiv:2608. 16622v1 Announce Type: cross Abstract: Multimodal harmful meme detection is typically formulated as image--text harmfulness classification.
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...
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.
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.
arXiv:2606. 28953v1 Announce Type: cross Abstract: Poisoning attacks entail attackers intentionally tampering with training data.
The community has recently developed various training-time defenses to counter neural backdoors introduced through data poisoning. In light of the observation that a model learns poisonous samples responsible for the backdoor easier than benign samples, these approaches either use a fixed threshold of the training loss for splitting or iteratively learn a reference model as an oracle for identifying benign samples.
arXiv:2606. 02947v1 Announce Type: new Abstract: Supervised fine-tuning is the predominant approach for adapting autoregressive vision-language models to downstream tasks.
Large language models (LLMs) are rigorously aligned to refuse harmful requests, a process that inherently cultivates a latent capacity to evaluate and recognize unsafe content. In this work, we reveal that this advanced safety awareness inadvertently introduces a fatal vulnerability.
arXiv:2606. 16808v1 Announce Type: new Abstract: While Large Reasoning Models (LRMs) excel at complex tasks, they remain highly vulnerable to sophisticated jailbreaks and direct harmful queries.
arXiv:2509.06896v3 Announce Type: replace Abstract: Targeted data poisoning attacks manipulate model predictions on specific test samples by injecting malicious data into training. Yet existing evalu...