Taxonomy of Risks posed by Language Models

By Laura Weidinger, Jonathan Uesato, Maribeth Rauh, Conor Griffin, Po-Sen Huang, John F. J. Mellor, Amelia Glaese, Myra Cheng, Borja Balle, Atoosa Kasirzadeh and 13 more. 2022

1,066 citations recorded by the source corpus.

Responsible innovation on large-scale Language Models (LMs) requires foresight into and in-depth understanding of the risks these models may pose. This paper develops a comprehensive taxonomy of ethical and social risks associated with LMs. We identify twenty-one risks, drawing on expertise and literature from computer science, linguistics, and the social sciences. We situate these risks in our taxonomy of six risk areas: I. Discrimination, Hate speech and Exclusion, II. Information Hazards, III. Misinformation Harms, IV. Malicious Uses, V. Human-Computer Interaction Harms, and VI. Environmental and Socioeconomic harms. For risks that have already been observed in LMs, the causal mechanism leading to harm, evidence of the risk, and approaches to risk mitigation are discussed. We further describe and analyse risks that have not yet been observed but are anticipated based on assessments of other language technologies, and situate these in the same taxonomy. We underscore that it is the responsibility of organizations to engage with the mitigations we discuss throughout the paper. We close by highlighting challenges and directions for further research on risk evaluation and mitigation with the goal of ensuring that language models are developed responsibly.

Subfields: Artificial Intelligence, Safety Research, Safety, Robustness & Interpretability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Safety Research

In Safety, Robustness & Interpretability

Citations in the corpus

Scaling Instruction-Finetuned Language Models (2022, background): “Finally, we note that our evaluations focus on only a subset of potential risks (Weidinger et al., 2022), and that our evaluations focus on model outputs without considering the wider sociotechnical context in which instruction-finetuned language models exist (Shelby et al.”
Foundation Models and Fair Use (2023, background): “Weidinger et al. (2022) taxonomize the risks of language models, noting copyright infringement and effects on creative economies.”
GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models (2023, background): “…of automation technologies, including LLMs, has previously been linked to heightened economic disparity and labor disruption, which may give rise to adverse downstream eﬀects (Acemoglu and Restrepo, 2022a; Acemoglu, 2002; Moll et al., 2021; Klinova and Korinek, 2021; Weidinger et al., 2021, 2022).”
LLM Censorship: A Machine Learning Challenge or a Computer Security Problem? (2023, background): “These risks encompass a wide range of issues, such as social engineering and data exﬁltration [26, 51], necessitating the development of methods to mitigate such risks by regulating LLM outputs.”
Generative artificial intelligence (2023, background): “Such issues, combined with low data quality and bias, provide a preview of potential social and ethical harms that may result in discrimination, exclusion, toxicity, and information hazards (Weidinger et al., 2022).”
Harms from Increasingly Agentic Algorithmic Systems (2023, background): “197], the generation of toxic language [7, 76], and informational harms [42, 99, 119, 193].”
Towards Measuring the Representation of Subjective Global Opinions in Language Models (2023, positive): “If a language model disproportionately represents certain opinions, it risks imposing potentially undesirable effects such as promoting hegemonic worldviews and homogenizing people’s perspectives and beliefs [7, 54, 79, 80, 16, 12, 23, 40].”
AlexaTM 20B: Few-Shot Learning Using a Large-Scale Multilingual Seq2Seq Model (2022, background): “Previous work has shown that large-scale pre-trained language models (LMs) can demonstrate undesirable biases (Sheng et al., 2021; Kurita et al.; Dev et al., 2020; Liang et al., 2021; Weidinger et al., 2022).”
HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal (2024, positive): “These categories were generating by referencing acceptable use policies at OpenAI, Anthropic, Meta, and Inflection AI, as well as literature on the potential for malicious use of LLMs (Weidinger et al., 2022; Hendrycks et al., 2023).”
mGPT: Few-Shot Learners Go Multilingual (2022, background): “LMs can reflect historical prejudices against disadvantaged social groups and reproduce harmful stereotypes about gender, race, religion, or sexual orientation (Weidinger et al., 2022).”
The ethics of ChatGPT in medicine and healthcare: a systematic review on Large Language Models (LLMs) (2024, background): “19,20 The healthcare and medical field, being particularly sensitive and heavily regulated, is notably susceptible to ethical dilemmas.”
Language Models Represent Space and Time (2023, background): “Better understanding of if and how LLMs model the world is critical for reasoning about the robustness, fairness, and safety of current and future AI systems (Bender et al., 2021; Weidinger et al., 2022; Bommasani et al., 2021; Hendrycks et al., 2023; Ngo et al., 2023).”
Co-Writing with Opinionated Language Models Affects Users’ Views (2023, background): “However, more cautious voices have also warned about the ethical and social risks of harm from large language models [100, 101], ranging from discrimination and exclusion [23, 54, 83] to misinformation [67, 75, 91, 106] and environmental [96] and socioeconomic harms [14].”