This page is still being polished. If you have thoughts, please share them via the feedback form.
Data on this page is preliminary and may change. Please do not share or cite these figures publicly.
Techniques to remove, bound, or modify learned model capabilities post-training.
Also in Model
Current approaches to enhancing the safety of LLMs exhibit notable limitations and inefficiencies, underscoring the urgent need for more robust and effective solutions
For instance, a study investigating the use of DPO to reduce toxicity (Lee et al., 2024a) has found that the alignment algorithm does not completely eliminate toxic content but instead circumvent it by bypassing sensitive neural regions. Additionally, model editing is inefficient either, as it requires manual, instance-by-instance modifications that only affect specific pieces of information without altering related content (Cohen et al., 2024; Zhong et al., 2023; Qin et al., 2024). Similarly, machine unlearning—designed to enable models to forget harmful information—has demonstrated instability. Experimental evidence suggests that even after harmful data is ostensibly erased, a machine-unlearned LLM may still recall private data, and with few-shot fine-tuning, it can revert to its previous state, recovering the erased information (Łucki et al., 2024; Lynch et al., 2024). Incomplete forgetting exacerbates the issue, as models can still leak private information when prompts are altered. Therefore, more effective methods are necessary. To meet the increasing safety demands of LLMs, a universal, unified mechanism capable of permanently eliminating unsafe information is imperative.
Future directions
Value Misalignment
99.9 OtherValue Misalignment > Mitigating social bias
1 AI SystemValue Misalignment > Privacy protection
1 AI SystemValue Misalignment > Methods for mitigating toxicity
1 AI SystemValue Misalignment > Methods for mitigating LLM amorality
1 AI SystemRobustness to attack
1 AI SystemLarge Language Model Safety: A Holistic Survey
Shi, Dan; Shen, Tianhao; Huang, Yufei; Li, Zhigen; Leng, Yongqi; Jin, Renren; Liu, Chuang; Wu, Xinwei; Guo, Zishan; Yu, Linhao; Shi, Ling; Jiang, Bojian; Xiong, Deyi (2024)
The rapid development and deployment of large language models (LLMs) have introduced a new frontier in artificial intelligence, marked by unprecedented capabilities in natural language understanding and generation. However, the increasing integration of these models into critical applications raises substantial safety concerns, necessitating a thorough examination of their potential risks and associated mitigation strategies. This survey provides a comprehensive overview of the current landscape of LLM safety, covering four major categories: value misalignment, robustness to adversarial attacks, misuse, and autonomous AI risks. In addition to the comprehensive review of the mitigation methodologies and evaluation resources on these four aspects, we further explore four topics related to LLM safety: the safety implications of LLM agents, the role of interpretability in enhancing LLM safety, the technology roadmaps proposed and abided by a list of AI companies and institutes for LLM safety, and AI governance aimed at LLM safety with discussions on international cooperation, policy proposals, and prospective regulatory directions. Our findings underscore the necessity for a proactive, multifaceted approach to LLM safety, emphasizing the integration of technical solutions, ethical considerations, and robust governance frameworks. This survey is intended to serve as a foundational resource for academy researchers, industry practitioners, and policymakers, offering insights into the challenges and opportunities associated with the safe integration of LLMs into society. Ultimately, it seeks to contribute to the safe and beneficial development of LLMs, aligning with the overarching goal of harnessing AI for societal advancement and well-being. A curated list of related papers has been publicly available at https://github.com/tjunlp-lab/Awesome-LLM-Safety-Papers.
Build and Use Model
Training, fine-tuning, and integrating the AI model
Developer
Entity that creates, trains, or modifies the AI system
Manage
Prioritising, responding to, and mitigating AI risks
Primary
7 AI System Safety, Failures & Limitations