The LMVD-ID is an internal research identifier, not an official CVE identifier.
Adversarial Unlearning Bypass
Large Language Models (LLMs) employing gradient-ascent based unlearning methods are vulnerable to a dynamic unlearning attack (DUA). DUA leverages optimized adversarial suffixes appended to prompts, reintroducing…
Paper-evaluated models(3)
- Llama 2 7B Chat
- Llama 3 8B Instruct
- Llama 3.1 8B Instruct
Description
Large Language Models (LLMs) employing gradient-ascent based unlearning methods are vulnerable to a dynamic unlearning attack (DUA). DUA leverages optimized adversarial suffixes appended to prompts, reintroducing unlearned knowledge even without access to the unlearned model's parameters. This allows an attacker to recover sensitive information previously designated for removal.
Examples
See the paper for specific examples of adversarial suffixes and their impact on unlearning different knowledge targets across various scenarios. The paper demonstrates successful retrieval of unlearned knowledge in 55.2% of tested cases, even without access to the unlearned model.
Impact
Successful exploitation of this vulnerability leads to the unintended disclosure of sensitive information previously removed from an LLM through unlearning. This compromises data privacy and confidentiality, violating the "right to be forgotten." The recovered knowledge can be used for malicious purposes such as reputational damage, intellectual property theft, or identity theft.
Affected Systems
Large Language Models (LLMs) utilizing gradient-ascent-based unlearning techniques, specifically those vulnerable to adversarial prompt engineering. The paper shows vulnerability in Llama-3-8B-Instruct models.
Mitigation Steps
- Implement the Latent Adversarial Unlearning (LAU) framework to enhance the robustness of the unlearning process.
- Integrate techniques like adversarial training during the unlearning phase to make the model more resistant to adversarial queries.
- Develop and deploy robust detection mechanisms to identify and filter malicious prompts attempting to recover unlearned knowledge. Monitor model behavior for unexpected outputs related to unlearned topics.
- Regularly update and retrain LLMs using improved unlearning methods to minimize vulnerabilities.
Research context and confidence
- Evidence and verification
- Paper-reported; independent reproduction is not documented.
- Primary research source linked.
- Severity
- Not rated by this catalog.
- Source and publication type
- arXiv · Research preprint.
- Peer-review status is not provided by this source.
- Author and publication status
- Author metadata is not stored; see the primary paper.
- Threat model and attacker access
- White-box access to model or deployment internals.
- Related deployment categories
- No related deployment category is classified.
- Taxonomy labels only; paper-specific deployment prerequisites are not inferred.
- Affected systems
- Large Language Models (LLMs) utilizing gradient-ascent-based unlearning techniques, specifically those vulnerable to adversarial prompt engineering. The paper shows vulnerability in Llama-3-8B-Instruct models.
Research Paper
Towards robust knowledge unlearning: An adversarial framework for assessing and improving unlearning robustness in large language models
Primary source: arXiv. Findings are reported by the cited research and have not been independently verified.
View PaperEvidence
This entry is based on a primary research source. Its findings are paper-reported; independent reproduction and verification are not claimed.
https://arxiv.org/abs/2408.10682Related research
- Parametric Model Unalignment
Published October 1, 2023 · model-layer, jailbreak, poisoning
- Voice Agent Behavioral Bypass
Published February 1, 2026 · model-layer, application-layer, injection
- Prompt Injection Alignment Bypass
Published September 1, 2025 · prompt-layer, model-layer, application-layer