Protecting society from AI misuse: when are restrictions on capabilities warranted?
Open accessThe DOI matches its Crossref record; 9 of the 9 references it cites were found in a registry or an index.
Abstract
Artificial intelligence (AI) systems will increasingly be used to cause harm as they grow more capable. In fact, AI systems are already starting to help automate fraudulent activities, violate human rights, create harmful fake images, and identify dangerous toxins. To prevent some misuses of AI, we argue that targeted interventions on certain capabilities will be warranted. These restrictions may include controlling who can access certain types of AI models, what they can be used for, whether outputs are filtered or can be traced back to their user, and the resources needed to develop them. We also contend that new restrictions on non-AI capabilities needed to cause harm will be required. For example, concerns about AI-enabled bioweapon acquisition have motivated efforts to introduce DNA synthesis screening. Though capability restrictions risk reducing use more than misuse (resulting in an unfavorable Misuse–Use Tradeoff), we argue that interventions on capabilities are warranted in some circumstances when other interventions are insufficient, the potential harm from misuse is high, and there are targeted interventions. We provide a taxonomy of interventions that can reduce AI misuse, focusing on the specific steps required for a misuse to cause harm (the Misuse Chain), and a framework to determine if an intervention is warranted. We exemplify our framework to three examples: predicting novel toxins, creating harmful images, and automating spear phishing campaigns.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Gorwa, R., Binns, R., & Katzenbach, C. (2020) · Big Data & Society · doi:10.1177/2053951719897945 · Cited by 1,097 · cited 1×
“Various forms of fingerprinting or hash matching (Cam-bridge Consultants 2019; Gorwa et al. 2020) are now used for identifying, removing, or reducing the spread of prohibited content (for example, copyrighted material or known child sexual abuse images) on social media platforms.”
The Malicious Use of Artificial Intelligence: Forecasting, Prevention, and Mitigation
Brundage, M., Avin, S., Clark, J., Toner, H., Eckersley, P., Garfinkel, B., Dafoe, A., Scharre, P., Zeitzoff, T., Filar, B., et al. (2018) · Apollo - University of Cambridge Repository · doi:10.17863/cam.22520 · Cited by 849
Bai, Y., Kadavath, S., Kundu, S., Askell, A., Kernion, J., Jones, A., Chen, A., Goldie, A., Mirhoseini, A., & Kaplan, J. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2212.08073 · Cited by 326 · cited 1×
“Finally, the content of propaganda campaigns may also change as messages can be made more credible if models are fine-tuned to mimic effective propagandists (Goldstein et al. 2023).”
Predictability and Surprise in Large Generative Models
Ganguli, D., Hernandez, D., Lovitt, L., DasSarma, N., Henighan, T., Jones, A., Joseph, N., Kernion, J., Mann, B., Askell, A., et al. (2022) · arXiv · doi:10.48550/arxiv.2202.07785 · Cited by 324
Toward Trustworthy AI Development: Mechanisms for Supporting Verifiable Claims
Brundage, M., Avin, S., Wang, J., Belfield, H., Krueger, G., Hadfield, G., Khlaaf, H., Yang, J., Toner, H., Fong, R., et al. (2020) · arXiv · doi:10.48550/arxiv.2004.07213 · Cited by 306 · cited 1×
“Further, they can invite external red teams to scrutinize their models and provide assessments before model release or deployment [11].”
Self-citationPredictability and Surprise in Large Generative Models
Ganguli, D., Hernandez, D., Lovitt, L., Askell, A., Bai, Y., Chen, A., Conerly, T., Dassarma, N., Drain, D., Elhage, N., et al. (2022) · 2022 ACM Conference on Fairness Accountability and Transparency · doi:10.1145/3531146.3533229 · Cited by 207 · cited 1×
“However, we should not expect these assessments to be fully comprehensive, as the uses to which AI systems can be put – including misuses – can be difficult to foresee, especially for large self-supervised models [26].”
BackgroundDevising and Detecting Phishing Emails Using Large Language Models
Heiding, F., Schneier, B., Vishwanath, A., Bernstein, J., & Park, P. S. (2024) · IEEE Access · doi:10.1109/access.2024.3375882 · Cited by 123
The Gradient of Generative AI Release: Methods and Considerations
Solaiman, I. (2023) · 2023 ACM Conference on Fairness, Accountability, and Transparency · doi:10.1145/3593013.3593981 · Cited by 90
How does the offense-defense balance scale?
Garfinkel, B., & Dafoe, A. (2019) · Journal of Strategic Studies · doi:10.1080/01402390.2019.1631810 · Cited by 73
Cited by
Yigit, Y., Ferrag, M. A., Ghanem, M. C., Sarker, I. H., Maglaras, L. A., Chrysoulas, C., Moradpoor, N., Tihanyi, N., & Janicke, H. (2025) · Sensors · doi:10.3390/s25061666 · Cited by 69
“Microsoft and OpenAI have reported early examples of AI being exploited by state-backed threat actors [124].”
Learning About AI: A Systematic Review of Reviews on AI Literacy
Zhang, S., Ganapathy Prasad, P., & Schroeder, N. L. (2025) · Journal of Educational Computing Research · doi:10.1177/07356331251342081 · Cited by 62
“This observation leads to another concern: misuse (Anderljung & Hazell, 2023; Veluru, 2024) or over-reliance on AI (Zhai et al., 2024).”
How to design an AI ethics board
Schuett, J., Reuel, A.-K., & Carlier, A. (2025) · AI and Ethics · doi:10.1007/s43681-023-00409-y · Cited by 46
Seger, E., Dreksler, N., Moulange, R., Dardaman, E., Schuett, J., Wei, K., Winter, C., Arnold, M., Ó hÉigeartaigh, S., Korinek, A., et al. (2023) · doi:10.2139/ssrn.4596436 · Cited by 31
“Potential epistemic, social and political consequences of foundation model misuse include the following [85, 86].”
BackgroundSelf-citation
Metadata
- Authors
- Anderljung, M., Hazell, J., & von Knebel, M.
- Venue
- AI & SOCIETY
- Published
- 2024-12-10
- Volume
- 40
- Issue
- 5
- Pages
- 3841-3857
- ISSN
- 0951-5666 (ISSN-L)
- 1435-5655
- OpenAlex
- W4405241549
- Cited by
- 45
- References
- 9
- FWCI
- 17.75
- Type
- journal article
- Language
- en
- Open access
- hybrid
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Protecting society from AI misuse: when are restrictions on capabilities warranted? https://app.cochat.ai/papers/10.1007/s00146-024-02130-8
- The paper
- Anderljung, M., Hazell, J., & von Knebel, M. (2025). Protecting society from AI misuse: when are restrictions on capabilities warranted? AI & SOCIETY. https://doi.org/10.1007/s00146-024-02130-8
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.