Pith. sign in

REVIEW 7 cited by

Trojan Activation Attack: Red-Teaming Large Language Models using Activation Steering for Safety-Alignment

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2311.09433 v3 pith:TH6SPZXZ submitted 2023-11-15 cs.CR cs.AIcs.CL

classification cs.CRcs.AIcs.CL
keywords modelsactivationattackllmsalignmentmakingsafetysteering
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

To ensure AI safety, instruction-tuned Large Language Models (LLMs) are specifically trained to ensure alignment, which refers to making models behave in accordance with human intentions. While these models have demonstrated commendable results on various safety benchmarks, the vulnerability of their safety alignment has not been extensively studied. This is particularly troubling given the potential harm that LLMs can inflict. Existing attack methods on LLMs often rely on poisoned training data or the injection of malicious prompts. These approaches compromise the stealthiness and generalizability of the attacks, making them susceptible to detection. Additionally, these models often demand substantial computational resources for implementation, making them less practical for real-world applications. In this work, we study a different attack scenario, called Trojan Activation Attack (TA^2), which injects trojan steering vectors into the activation layers of LLMs. These malicious steering vectors can be triggered at inference time to steer the models toward attacker-desired behaviors by manipulating their activations. Our experiment results on four primary alignment tasks show that TA^2 is highly effective and adds little or no overhead to attack efficiency. Additionally, we discuss potential countermeasures against such activation attacks.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 7 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Paladin: Defending LLM-enabled Phishing Emails with a New Trigger-Tag Paradigm

    cs.CR 2025-09 conditional novelty 7.0 of 10

    A trigger-tag watermark embedded by fine-tuning lets modified LLMs mark their own phishing outputs for cheap detection.

  2. Steering Protein Language Models

    q-bio.BM 2025-07 reject novelty 6.0 of 10

    Activation steering can guide protein language models to generate and optimize sequences with higher predicted thermostability, solubility, or GFP brightness, but only in surrogate-based evaluation.

  3. Turning the Spell Around: Lightweight Alignment Amplification via Rank-One Safety Injection

    cs.CL 2025-08 conditional novelty 5.0 of 10

    ROSI bakes the refusal direction into a model's weight matrices via a rank-one update, raising refusal and jailbreak robustness with minimal measured utility cost.

  4. Layer-Wise Perturbations via Sparse Autoencoders for Adversarial Text Generation

    cs.CL 2025-08 reject novelty 5.0 of 10

    Sparse autoencoder activation perturbation (SFPF) applied on top of existing jailbreak prompts raises attack success rate on Qwen3-32B, but with no defense evaluation and weak reproducibility.

  5. Security Concerns for Large Language Models: A Survey

    cs.CR 2025-05 conditional novelty 5.0 of 10

    A survey that classifies LLM security threats and argues that intrinsic agentic risks, such as scheming, are underappreciated and poorly defended.

  6. Probing the Robustness of Large Language Models Safety to Latent Perturbations

    cs.LG 2025-06 conditional novelty 4.0 of 10

    Randomized noise injected into hidden layers bypasses safety refusals in 12 open LLMs, and layer-wise adversarial training on the resulting benchmark reduces the attack's success.

  7. Breaking to Build: A Threat Model of Prompt-Based Attacks for Securing LLMs

    cs.CL 2025-09 conditional novelty 3.0 of 10

    A survey categorizing prompt-based attacks on LLMs into four classes and proposing aspirational goals of un-distillable, un-finetunable, and un-editable models.

Pith tools