The DOI matches its Crossref record; 114 of the 176 references it cites were found in a registry or an index.
Regulating ChatGPT and other Large Generative AI Models
Open accessAbstract
Large generative AI models (LGAIMs), such as ChatGPT, GPT-4 or Stable Diffusion, are rapidly transforming the way we communicate, illustrate, and create. However, AI regulation, in the EU and beyond, has primarily focused on conventional AI models, not LGAIMs. This paper will situate these new generative models in the current debate on trustworthy AI regulation, and ask how the law can be tailored to their capabilities. After laying technical foundations, the legal part of the paper proceeds in four steps, covering (1) direct regulation, (2) data protection, (3) content moderation, and (4) policy proposals. It suggests a novel terminology to capture the AI value chain in LGAIM settings by differentiating between LGAIM developers, deployers, professional and non-professional users, as well as recipients of LGAIM output. We tailor regulatory duties to these different actors along the value chain and suggest strategies to ensure that LGAIMs are trustworthy and deployed for the benefit of society at large. Rules in the AI Act and other direct regulation must match the specificities of pre-trained models. The paper argues for three layers of obligations concerning LGAIMs (minimum standards for all LGAIMs; high-risk obligations for high-risk use cases; collaborations along the AI value chain). In general, regulation should focus on concrete high-risk applications, and not the pre-trained model itself, and should include (i) obligations regarding transparency and (ii) risk management. Non-discrimination provisions (iii) may, however, apply to LGAIM developers. Lastly, (iv) the core of the DSA's content moderation rules should be expanded to cover LGAIMs. This includes notice and action mechanisms, and trusted flaggers.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“For example, improperly secured LLMs could enable bad actors to generate fake patient data or insurance claims [62].”
From Generative AI in medical practice: in-depth exploration of privacy and security challenges (2024) · cites this paper as [62]
Experimental — accuracy not yet measured
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
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404 · cited 2×
“Transparent LGAIMs connected to (good-old) knowledge bases, such as AtMan [43], present a promising way forward in this direction.”
BackgroundBERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372 · cited 4×
“LGAIMs are usually trained with several billion, if not hundreds of billions, parameters [43, 44], requiring large amounts of training data and computing power [45].”
BackgroundInfluentialLearning Transferable Visual Models From Natural Language Supervision
Radford, A., Kim, J. W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., et al. (2021) · arXiv · doi:10.48550/arxiv.2103.00020 · Cited by 15,293 · cited 1×
“OpenAI’s “CLIP” image classifier, for example, was built using a set of 400 million image-text pairs [46].”
MethodLewis, M., Liu, Y., Goyal, N., Ghazvininejad, M., Mohamed, A., Levy, O., Stoyanov, V., & Zettlemoyer, L. (2019) · arXiv · doi:10.48550/arxiv.1910.13461 · Cited by 8,816 · cited 1×
“Although the emergence of these models in recent years constitutes a significant technical advance (for foundations, see [39-43]), they harness, to a reasonable extent, existing technologies in a vastly increased scale and scope.”
BackgroundLiu, P., Yuan, W., Fu, J., Jiang, Z., Hayashi, H., & Neubig, G. (2023) · ACM Computing Surveys · doi:10.1145/3560815 · Cited by 4,361
Liu, P., Yuan, W., Fu, J., Jiang, Z., Hayashi, H., & Neubig, G. (2021) · arXiv · doi:10.48550/arxiv.2107.13586 · Cited by 4,352
Training language models to follow instructions with human feedback
Ouyang, L., Wu, J., Jiang, X., Almeida, D., Wainwright, C. L., Mishkin, P., Zhang, C., Agarwal, S., Slama, K., Ray, A., et al. (2022) · arXiv · doi:10.48550/arxiv.2203.02155 · Cited by 4,208
Counterfactual Explanations without Opening the Black Box: Automated Decisions and the GDPR
Wachter, S., Mittelstadt, B., & Russell, C. (2017) · arXiv.org · doi:10.48550/arxiv.1711.00399 · Cited by 3,582
Training Language Models to Follow Instructions with Human Feedback
Ouyang, L., Wu, J., Jiang, X., Almeida, D., Wainwright, C., Mishkin, P., Zhang, C., Agarwal, S., Slama, K., Ray, A., et al. (2022) · Advances in Neural Information Processing Systems 35 · doi:10.52202/068431-2011 · Cited by 3,117 · cited 1×
“2 LGAIMs can often digest human text input [57, 58] and produce an output (text; image; audio; video) based on it.”
Background
Research works printed but not linked: 44
- Marcus, G. A Skeptical Take on the A.I. Revolution. The Ezra Klein Show, The New York Times, 2023.
- Beuth, P. Wie sich ChatGPT mit Worten hacken lässt. Der Spiegel, 2023.
- Satariano, A. and Mozur, P. The People Onscreen Are Fake. The Disinformation Is Real., 2023.
- Edwards, L. Regulating AI in Europe: four problems and four solutions (2022), 2022.
- Hacker, P., Engel, A. and List, T. Understanding and regulating ChatGPT, and other large generative AI models. 2023.
- Heikkilä, M. The EU wants to regulate your favorite AI tools. 2023.
- Chee, F. Y. and Mukherjee, S. Exclusive: ChatGPT in spotlight as EU's Breton bats for tougher AI rules. Reuters, 2023.
- Smith, B. Meeting the AI moment: advancing the future through responsible AI. 2023.
- Lieu, T. I'm a Congressman Who Codes. A.I. Freaks Me Out., 2023.
- Douek, E. Content Moderation as Systems Thinking. Harv. L. Rev., 136 (2022), 526.
- 10.1007/978-3-030-95220-4_4 · DOI
- Meyer, P. ChatGPT: How Does It Work Internally? , 2022.
- Sassoli, M. Autonomous weapons and international humanitarian law: Advantages, open technical questions and legal issues to be clarified. International Law Studies, 90, 1 (2014), 1.
- Radford, A., Narasimhan, K., Salimans, T. and Sutskever, I. Improving language understanding by generative pre-training (2018).
- Bienert, J. and Klös, H.-P. Große KI-Modelle als Basis für Forschung und wirtschaftliche Entwicklung. IW-Kurzbericht, 2022.
- OECD Measuring the Environmental Impacts of AI Compute and Applications: The AI Footprint. 2022.
- ACM, T. P. C. ACM TechBrief: Computing and Climate Change. 2021.
- 10.1016/j.oneear.2021.05.018 · DOI
- Ananthaswamy, A. The Physics Principle That Inspired Modern AI Art. 2023.
- Luccioni, A. S. and Viviano, J. D. What's in the Box? A Preliminary Analysis of Undesirable Content in the Common Crawl Corpus. arXiv preprint arXiv:2105.02732 (2021). · arXiv
- Perrigo, B. OpenAI Used Kenyan Workers on Less Than $2 Per Hour to Make ChatGPT Less Toxic. 2023.
- Bertuzzi, L. Leading MEPs exclude general-purpose AI from high-risk categories – for now. 2022.
- Bertuzzi, L. AI Act: EU Parliament's crunch time on high-risk categorisation, prohibited practices. 2023.
- Bertuzzi, L. AI Act: MEPs close in on rules for general purpose AI, foundation models. 2023.
- Bertuzzi, L. MEPs seal the deal on Artificial Intelligence Act. 2023.
- Liang, P., Bommasani, R., Creel, K. and Reich, R. The time is now to develop community norms for the release of foundation models. 2022.
- Bornstein, M., Appenzeller, G. and Casado, M. Who Owns the Generative AI Platform? , 2023.
- Micklitz, H.-W., Stuyck, J., Terryn, E. and School, I. C. Consumer law. Hart London, 2010.
- Meyers, J. M. Artificial intelligence and trade secrets. Landslide, 11 (2018), 17.
- Bertuzzi, L. Leading EU lawmakers propose obligations for General Purpose AI. 2023.
- Calvin, N. and Leung, J. Who owns artificial intelligence? A preliminary analysis of corporate intellectual property strategies and why they matter. Future of Humanity Institute, February (2020).
- Deeks, A. The judicial demand for explainable artificial intelligence. Columbia Law Review, 119, 7 (2019), 1829-1850.
- McKown, J. R. Discovery of Trade Secrets. Santa Clara Computer & High Tech. LJ, 10 (1994), 35.
- Subrin, S. N. Discovery in Global Perspective: Are We Nuts. DePaul L. Rev., 52 (2002), 299.
- Daniel, P. F. Protecting Trade Secrets from Discovery. Tort & Ins. LJ, 30 (1994), 1033.
- Wachter, S. Affinity profiling and discrimination by association in online behavioral advertising. Berkeley Tech. LJ, 35 (2020), 367.
- Kiela, D., Firooz, H., Mohan, A., Goswami, V., Singh, A., Fitzpatrick, C. A., Bull, P., Lipstein, G., Nelli, T. and Zhu, R. The hateful memes challenge: Competition report. PMLR, 2021.
- Seeha, S. Prompt Engineering and Zero-Shot/Few-Shot Learning [Guide]. 2022.
- European, C., Directorate-General for Communications Networks, C., Technology, Hoboken, J., Quintais, J., Poort, J. and Eijk, N. Hosting intermediary services and illegal content online : an analysis of the scope of article 14 ECD in light of developments in the online service landscape : final report. Publications Office, 2019.
- 10.1017/CBO9780511676161 · DOI
- Gerdemann, S. and Spindler, G. Das Gesetz über digitale Dienste (Digital Services Act) (Part 2). Gewerblicher Rechtsschutz und Urheberrecht (2023), 115-125.
- Bertuzzi, L. AI Act: MEPs want fundamental rights assessments, obligations for high-risk users. 2023.
- Crootof, R. Artificial intelligence research needs responsible publication norms. Lawfare Blog (2019).
- Hoffmann-Riem, W. Innovation und Recht-Recht und Innovation: Recht im Ensemble seiner Kontexte. Mohr Siebeck, 2016.
Cited by
Sorted by citations
Items 101–200 of 493
Autonomous Agents in Software Development: A Vision Paper
Rasheed, Z., Waseem, M., Sami, M., Kemell, K., Ahmad, A., Nguyen‐Duc, A., Systä, K., & Abrahamsson, P. (2025) · Lecture notes in business information processing · doi:10.1007/978-3-031-72781-8_2 · Cited by 18
“Nowadays, generative AI has been utilized in various fields, such as NLP, computer vision, and image and video generation [15].”
Wang, Y., Sauka, K., & Situmeang, F. (2025) · Journal of Consumer Marketing · doi:10.1108/jcm-04-2024-6806 · Cited by 67
A Comprehensive Survey on Generative AI for Metaverse: Enabling Immersive Experience
Chamola, V., Sai, S., Bhargava, A., Sahu, A., Jiang, W., Xiong, Z., Niyato, D. T., & Hussain, A. (2024) · Cognitive Computation · doi:10.1007/s12559-024-10342-9 · Cited by 42
“Different countries or regions may have other guidelines and regulations [293], and adhering to each can be a challenge for Generative AI companies, primarily since the Metaverse is intended to be used by people worldwide.”
Comparative Analysis of Generative AI Risks in the Public Sector
Navarro, M. A. B., Mondragón, M. I. R., & Han, S. H. (2024) · Annual International Conference on Digital Government Research · doi:10.1145/3657054.3657125 · Cited by 48
“The landscape of artificial intelligence (AI) has experienced a mon-umental shift [1], in particular, with the advent and rapid diffusion of generative AI (GenAI) impacting all sectors worldwide [2], [3].”
BackgroundInfluentialAre Watermarks Bugs for Deepfake Detectors? Rethinking Proactive Forensics
Wu, X.-S., Liao, X., Ou, B., Liu, Y., & Qin, Z. (2024) · International Joint Conference on Artificial Intelligence · doi:10.48550/arxiv.2404.17867 · Cited by 36
“Large generative models like ChatGPT and Stable Diffusion are transforming the way we communicate, illustrate, and create [Hacker et al. , 2023].”
MethodModerating model marketplaces: platform governance puzzles for AI intermediaries
Gorwa, R., & Veale, M. (2024) · Law Innovation and Technology · doi:10.1080/17579961.2024.2388914 · Cited by 42
Humans are still better than ChatGPT: Case of the IEEEXtreme competition
Koubâa, A., Qureshi, B., Ammar, A., Khan, Z., Boulila, W., & Ghouti, L. (2023) · Heliyon · doi:10.1016/j.heliyon.2023.e21624 · Cited by 46
“Moreover, it can provide customer support to handle routine inquiries.”
BackgroundCharting the Growth and Structure of Early ChatGPT-Education Research: A Bibliometric Study
Watrianthos, R., Ahmad, S. T., & Muskhir, M. (2023) · Journal of Information Technology Education Innovations in Practice · doi:10.28945/5221 · Cited by 47
“For example, the emphasis on AI tools and generative models suggests optimism about leveraging ChatGPT ’ s capabilities to transform practices (Hacker et al., 2023).”
BackgroundHwang, A. H., Siy, J. O., Shelby, R. M., & Lentz, A. (2024) · Designing Interactive Systems Conference · doi:10.1145/3643834.3661555 · Cited by 34
“The general public has been increasingly involved in interrogating the potential of this emerging technology – including the potential risks it poses [20, 24, 26, 33, 43, 52, 95, 116].”
BackgroundEmbarrassingly Simple Text Watermarks
Sato, R., Takezawa, Y., Bao, H., Niwa, K., & Yamada, M. (2023) · arXiv.org · doi:10.48550/arxiv.2310.08920 · Cited by 35
Distinguishing Human Generated Text From ChatGPT Generated Text Using Machine LearningPreprint
Islam, N., Sutradhar, D., Noor, H., Raya, J. T., Maisha, M. T., & Farid, D. M. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2306.01761 · Cited by 40
“One of the biggest concerns is that, it can produce false news or spread misinformation [1] [2].”
BackgroundRisks and Opportunities of Open-Source Generative AI
Eiras, F., Petrov, A., Vidgen, B., Schroeder, C., Pizzati, F., Elkins, K., Mukhopadhyay, S., Bibi, A., Purewal, A., Botos, C., et al. (2024) · arXiv · doi:10.48550/arxiv.2405.08597
“A considerable number of opinion articles (LAION.ai, 2023; Hacker et al., 2023; Creative Commons, 2023) et al., 2023; Andreessen Horowitz, 2023a; Shrestha et al., 2023), while improving developments and facilitating further research on the topic (Kasneci et al., 2023; ArsTechnica, 2023; Spirling,…”
BackgroundSafety and Fairness for Content Moderation in Generative Models
Hao, S., Kumar, P., Laszlo, S., Poddar, S., Radharapu, B., & Shelby, R. (2023) · arXiv.org · doi:10.48550/arxiv.2306.06135 · Cited by 34
“[28], situates harms from generative AI systems in the context of the Digital Services Act, through which content moderation would play a central role.”
BackgroundMobilising new frontiers in digital transformation research: A problematization review
Ashrafi, A., Constantinides, P., Mehandjiev, N., & Thatcher, J. B. (2025) · Information Systems Journal · doi:10.1111/isj.12531 · Cited by 33
“Many antitrust investigations have focused on the anticompetitive practices of dominant organisations on third parties, because of their dual role as producers and orchestrators of generative AI solutions (Hacker et al., 2023).”
Chugh, R., Turnbull, D., Kutty, S., Sabrina, F., Rashid, M. M., Morshed, A., Azad, S., Kaisar, S., & Subramani, S. (2025) · Education and Information Technologies · doi:10.1007/s10639-025-13686-3 · Cited by 40
“The AI Act and related regulatory frameworks must consider the unique characteristics of trained models to address their implications in educational contexts (Hacker et al., 2023).”
Balayn, A., Yurrita, M., Rancourt, F., Casati, F., & Gadiraju, U. (2025) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3706598.3713787 · Cited by 37
“We focused on LLMs due to their timely relevance and concomitant supply chain governance challenges [55]: with the rise of LLMs, many small and medium-sized companies depend on foundation models provided by larger organizations and fine-tuned to their specific needs by intermediate organizations.”
Benign Samples Matter! Fine-tuning On Outlier Benign Samples Severely Breaks Safety
Guan, Z.-H., Hu, M.-X., Zhu, R.-H., Li, S., & Vullikanti, A. (2025) · International Conference on Machine Learning · doi:10.48550/arxiv.2505.06843 · Cited by 32
“(2) Fine-tuning-stage strategies (Mukhoti et al., 2023; Qi et al., 2024; Bianchi et al., 2023; Huang et al., 2024d; Lyu et al., 2024; Hacker et al., 2023; Choi et al., 2024), and (3) Post-deployment strategies (Huang et al., 2024a; Du et al., 2024b).”
Redefining Virtual Assistants in Health Care: The Future With Large Language Models
Sezgın, E. (2024) · Journal of Medical Internet Research · doi:10.2196/53225 · Cited by 49
“…that the development and integration of VAs are guided by a balanced perspective that considers ethical guidelines and regulatory oversight [25,26], governance principles [27,28], privacy and safety measures [29], feasibility, efficacy, and patient-centric approaches and assessment methods…”
Wang, X., Hong, Y., & He, X. (2024) · IET Blockchain · doi:10.1049/blc2.12076 · Cited by 38
CodePori: Large Scale Model for Autonomous Software Development by Using Multi-Agents
Rasheed, Z., Waseem, M., Saari, M., Systä, K., & Abrahamsson, P. (2024) · arXiv.org · doi:10.48550/arxiv.2402.01411 · Cited by 22
The technology triad: disruptive AI, regulatory gaps and value change
Hopster, J. K. G., & Maas, M. M. (2023) · AI and Ethics · doi:10.1007/s43681-023-00305-5 · Cited by 42
Parikh, N. A. (2023) · Advances in Business Strategy and Competitive Advantage · doi:10.4018/979-8-3693-1634-4.ch016 · Cited by 29
Position: Towards Bidirectional Human-AI Alignment
Shen, H., Knearem, T., Ghosh, R., Alkiek, K., Krishna, K., Yachuan, L., Petridis, S., Peng, Y.-H., Qiwei, L., Si, C., et al. (2025) · neural information processing systems · doi:10.52202/085713-4186 · Cited by 89
“[104, 105, 130].”
Generative AI Models in Time-Varying Biomedical Data: Scoping Review
He, R., Sarwal, V., Qiu, X., Zhuang, Y., Zhang, L., Liu, Y., & Chiang, J. N. (2024) · Journal of Medical Internet Research · doi:10.2196/59792 · Cited by 36
Cha, S. (2024) · Humanities and Social Sciences Communications · doi:10.1057/s41599-024-03017-1 · Cited by 44
“Moreover, a system where humans can intervene in real-time is crucial for promptly correcting AI errors, minimizing the resulting damages (Hacker, 2023).”
InfluentialLarge Language Models "Ad Referendum": How Good Are They at Machine Translation in the Legal Domain?
Briva-Iglesias, V., Camargo, J., & Doğru, G. (2024) · MonTI : monographs in translation and interpreting · doi:10.48550/arxiv.2402.07681 · Cited by 29
Diffusing the Creator: Attributing Credit for Generative AI Outputs
Khosrowi, D., Finn, F., & Clark, E. (2023) · Proceedings of the AAAI/ACM Conference on AI Ethics and Society · doi:10.1145/3600211.3604716 · Cited by 28
“whose products access GAIs through APIs) and users of their general AI models (including the likes of ChatGPT and Stable Diffusion), rather than taking accountability for potential damages themselves [30, 61].”
Kaiser, K. N., Hughes, A. J., Yang, A. D., Turk, A. A., Mohanty, S., Gonzalez, A. A., Patzer, R. E., Bilimoria, K. Y., & Ellis, R. J. (2024) · Journal of Surgical Oncology · doi:10.1002/jso.27821 · Cited by 29
“29,30 34 The Office of the National Coordinator for Health Information Technology (ONC), Department of Health and Human Services (HHS) identified and addressed that transparency of how a model was designed, developed, trained, and evaluated to establish and improve provider trust with use of these…”
Frontier AI developers need an internal audit function
Schuett, J. (2024) · Risk Analysis · doi:10.1111/risa.17665 · Cited by 50
“…is about GPAIS (e.g., Barrett et al., 2023), foundation models (e.g., Kapoor et al., 2024; Partnership on AI, 2023; Seger, Dreksler et al., 2023; Vipra & Korinek, 2023), generative AI (e.g., Hacker et al., 2023; Longpre et al., 2024), or AGI (e.g., Koessler & Schuett, 2023; Schuett et al., 2023).”
Zou, M., & Zhang, L. (2025) · Law and Governance · doi:10.1017/cfl.2024.4 · Cited by 29
“The second pathway adopts a more comprehensive approach that encompasses all potential applications or scenarios (Hacker et al., 2023).”
AI Regulation in Europe: From the AI Act to Future Regulatory ChallengesPreprint
Hacker, P. (2023) · arXiv.org · doi:10.48550/arxiv.2310.04072 · Cited by 27
Self-citationCodepori: Large-Scale System for Autonomous Software Development Using Multi-Agent Technology
Rasheed, Z., Malik, S., Kemell, K.-K., Muhammad, W., Saari, M., Systä, K., & Abrahamsson, P. (2024) · doi:10.2139/ssrn.4979510 · Cited by 27
Artificial Intelligence and Privacy
Solove, D. J. (2024) · SSRN Electronic Journal · doi:10.2139/ssrn.4713111 · Cited by 33
Cordella, A., & Gualdi, F. (2024) · Government Information Quarterly · doi:10.1016/j.giq.2024.101982 · Cited by 37
Beyond phase-in: assessing impacts on disinformation of the EU Digital Services Act
Nannini, L., Bonel, E., Bassi, D., & Maggini, M. J. (2024) · AI and Ethics · doi:10.1007/s43681-024-00467-w · Cited by 38
“Provisions emphasize metaphorical notions of “ virality ” instead of granular technical factors like source credibility and message exposure frequencies psychologically enabling belief formation (Articles 26-27, Codes: NTD, RA) [63, 64, 78].”
InfluentialKodali, R. K., Upreti, Y. P., & Boppana, L. (2023) · IEEE International Conference on Humanoid, Nanotechnology, Information Technology, Communication and Control, Environment, and Management (HNICEM) · doi:10.1109/hnicem60674.2023.10589199 · Cited by 3
“Finally, collaborative efforts culminate in Governance of ChatGPT and Similar Large Generative AI Models [12], navigating the rapidly evolving landscape of LGAIMs, which hold revolutionary sway over communication, creativity, and various sectors.”
Exploring Generative Artificial Intelligence: A Taxonomy and Types
Strobel, G., Banh, L., Möller, F., & Schoormann, T. (2024) · Hawaii International Conference on System Sciences (HICSS) · doi:10.24251/hicss.2024.546 · Cited by 30
“Non-DL-based generative models such as Hidden Markov Models play a minor role in recent discussions on GAI and realistic data generation (Gm et al., 2020; Hacker et al., 2023).”
BackgroundZahid, I. A., Joudar, S. S., Albahri, A. S., Albahri, O. S., Alamoodi, A. H., Santamaría, J., & Alzubaidi, L. (2024) · Intelligent Systems with Applications · doi:10.1016/j.iswa.2024.200431 · Cited by 29
Invisible Saboteurs: Sycophantic LLMs Mislead Novices in Problem-Solving Tasks
Bo, J. Y., Kazemitabaar, M., Deng, M., Inzlicht, M., & Anderson, A. (2026) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3791365 · Cited by 33
“While obviously challenging (and not absolving model developers of their responsibility), systemic structures like educational institutions and regulatory bodies can also adapt their policies to meet the opportunities and challenges posed by Generative AI [33].”
Mental Models of Generative AI Chatbot Ecosystems
Wang, X.-Y., Wang, X.-Z., Park, S., & Yao, Y.-X. (2025) · International Conference on Intelligent User Interfaces · doi:10.1145/3708359.3712125 · Cited by 24
Generative artificial intelligence in fifth-generation education systems: A systematic review
Chakraborty, S. (2026) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2026.114463 · Cited by 65
“It is a frequently utilised tool in both personal and professional contexts due to its capacity to comprehend the context and give logical, contextually relevant solutions [3], [18], [24], [28], [29], [38], [39], [40], [42], [46], [55], [57], [58], [62].”
Li, Z. S., Arony, N. N., Awon, A. M., Damian, D., & Xu, B. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2406.17325 · Cited by 43
“” (P4) However, the privacy discussion is ongoing, where the development of stringent regulations is being discussed in various jurisdictions around the world [21, 102, 111] amidst calls for more regulation [17, 43, 46, 53, 83].”
BackgroundCybersecurity Issues in Generative AI
Oh, S., & Shon, T. (2023) · International Conference on Platform Technology and Service · doi:10.1109/platcon60102.2023.10255179 · Cited by 23
Parikh, N. (2023) · arXiv.org · doi:10.48550/arxiv.2306.04605 · Cited by 23
“These principles, along with a thorough understanding of legal and regulatory standards, such as the GDPR (2023) and AI Act (2023), can help guide the development and implementation of generative AI technologies (Hacker et al., 2023).”
BackgroundValueCompass: A Framework of Fundamental Values for Human-AI AlignmentPreprint
Shen, H., Knearem, T., Ghosh, R., Yang, Y.-J., Mitra, T., & Huang, Y. (2024) · arXiv.org · doi:10.48550/arxiv.2409.09586 · Cited by 23
“…about AI (e.g., AI ethics [14] and auditing [54]), human-AI collaboration (e.g., developing AI assistants [101] or tutors [58]), user interaction and experience design (e.g., participatory design of AI [102]), and social impact of AI (e.g., misinformation [8], AI regulatory and policy [37]).”
BackgroundChatGPT-generated versus human direct corrective feedback on L2 writing
Muñoz Muñoz, B. C., Nassaji, H., & Bello Carrillo, F. I. (2025) · System · doi:10.1016/j.system.2025.103805 · Cited by 22
Ethical and practical challenges of generative AI in healthcare and proposed solutions: a survey
Tung, T., Hasnaeen, S. M. N., & Zhao, X. (2025) · Frontiers in Digital Health · doi:10.3389/fdgth.2025.1692517 · Cited by 31
“Actors can use the data of patients, or even ask an AI system to generate them fake data, and try to make insurance claims (56).”
Torkamaan, H., Steinert, S., Pera, M. S., Kudina, O., Freire, S. K., Verma, H., Kelly, S., Sekwenz, M.-T., Yang, J., van Nunen, K., et al. (2024) · Behaviour and Information Technology · doi:10.1080/0144929x.2024.2431068 · Cited by 31
Sarigul, B., Schneider, F. M., & Utz, S. (2024) · International journal of human computer interactions · doi:10.1080/10447318.2024.2375797 · Cited by 22
“At the same time, there is evidence that these AI models often generate fabricated information (Hacker et al., 2023).”
Animation and Artificial Intelligence
Stark, L. (2024) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3630106.3658995 · Cited by 19
“Existing legal and regulatory frameworks seem likely to be necessary but not sufficient [40].”
BackgroundEdge Intelligence Optimization for Large Language Model Inference with Batching and Quantization
Zhang, X.-Y., Liu, J., Xiong, Z.-H., Huang, Y., Xie, G.-C., & Zhang, R. (2024) · IEEE Wireless Communications and Networking Conference · doi:10.1109/wcnc57260.2024.10571127 · Cited by 22
“In the past year, Generative AI (GAI) has revolutionized the AI field, automating content creation and liberating creators from time-consuming manual efforts [1].”
BackgroundA Survey on Human–AI Collaboration with Large Foundation Models
Vats, V., Nizam, M. B., Liu, M., Wang, Z., Ho, R. C., Prasad, M. S., Titterton, V., Malreddy, S. V., Aggarwal, R., Xu, Y., et al. (2026) · ACM Transactions on Intelligent Systems and Technology · doi:10.1145/3841472 · Cited by 73
“The diverse roles of humans, ranging from providing ethical oversight in the design phase to contributing to quality control and data labeling during model training and execution, are crucial for optimizing AI performance.”
Applying Large Language Models in Legal Translation: The State-of-the-Art
Bajcic, M. (2025) · Social Science Research Network · doi:10.2139/ssrn.5040587 · Cited by 22
Copyright in the Artificial Intelligence Act – A Primer
Peukert, A. (2024) · Social Science Research Network · doi:10.2139/ssrn.4771976 · Cited by 22
Correctness is not Faithfulness in Retrieval Augmented Generation Attributions
Wallat, J., Heuss, M., Rijke, M. D., & Anand, A. (2025) · Proceedings of the 2025 International ACM SIGIR Conference on Innovative Concepts and Theories in Information Retrieval (ICTIR) · doi:10.1145/3731120.3744592 · Cited by 21
Rethinking technology regulation in the age of AI risks
Currie, W. L., Leimeister, J., Schlagwein, D., & Willcocks, L. P. (2025) · Journal of Information and Technology · doi:10.1177/02683962251378815 · Cited by 21
A General Framework to Enhance Fine-tuning-based LLM Unlearning
Ren, J., Dai, Z.-W., Tang, X.-F., Liu, H., Zeng, J.-Y., Li, Z., Goutam, R., Wang, S.-H., Xing, Y., & He, Q. (2025) · Annual Meeting of the Association for Computational Linguistics · doi:10.48550/arxiv.2502.17823 · Cited by 21
“However, concerns have been raised regarding the use of such large-scale data, as it often includes copyrighted and privacy-protected data (Hacker et al., 2023; Lucchi, 2024).”
The effects of generative AI’s human-like competencies on clinical decision-making
Spatscheck, N., Schaschek, M., & Winkelmann, A. (2024) · Journal of Decision System · doi:10.1080/12460125.2024.2430731 · Cited by 28
Wang, Z., Shen, L., Kuang, E., Zhang, S., & Fan, M. (2024) · Designing Interactive Systems Conference · doi:10.1145/3643834.3660703 · Cited by 21
“Furthermore, previous studies have highlighted other challenges posed by AIGC, such as contentious intellectual property rights of AI-generated content [27], and privacy and security concerns associated with AIGC tool usage [14].”
BackgroundGhandour, A., Woodford, B. J., & Abusaimeh, H. (2024) · IEEE Access · doi:10.1109/access.2024.3394243 · Cited by 35
“However, protecting algorithms as trade secrets could conflict with the growing demand for transparency in AI, thereby posing an ethical dilemma [21].”
BackgroundAI Risk Categorization Decoded (AIR 2024)
Zeng, Y., Klyman, K., Zhou, A., Yang, Y., Pan, M.-Z., Jia, R.-X., Song, D., Liang, P., & Li, B. (2024) · AGI - Artificial General Intelligence - Robotics - Safety & Alignment · doi:10.70777/agi.v1i1.10603 · Cited by 20
Fernández-Llorca, D., Gómez, E., Sánchez, I., & Mazzini, G. (2024) · Artificial Intelligence and Law · doi:10.1007/s10506-024-09412-y · Cited by 20
Regulating Emerging Technologies: Prospective Sensemaking Through Abstraction And Elaboration
Seidel, S., Frick, C. J., & Brocke, J. V. (2024) · MIS Quarterly · doi:10.25300/misq/2024/18039 · Cited by 40
“Such was the case with Bitcoin and Ethereum, which raised questions about blockchain regulation (Finck, 2018; Kiviat, 2015) and have recently been featured in debates and political processes around regulating generative AI (Hacker et al., 2023).”
A survey on biomedical automatic text summarization with large language models
Huang, Z., Chen, X., Wang, Y., Huang, J., & Zhao, X. (2025) · Information Processing & Management · doi:10.1016/j.ipm.2025.104216 · Cited by 24
Kaiser, K. N., Hughes, A. J., Yang, A. D., Mohanty, S., Maatman, T. K., Gonzalez, A. A., Patzer, R. E., Bilimoria, K. Y., & Ellis, R. J. (2025) · Surgery · doi:10.1016/j.surg.2025.109267 · Cited by 18
Legal implications of using generative AI in the media
Bayer, J. (2024) · Information & Communications Technology Law · doi:10.1080/13600834.2024.2352694 · Cited by 19
“38 The special needs of vulnerable groups due to their age or disability should also be taken into account when the AI system is intended to interact with those groups as well.”
ChatGPT Role-play Dataset: Analysis of User Motives and Model Naturalness
Tao, Y.-F., Agrawal, A., Dombi, J., Sydorenko, T., & Lee, J.-I. (2024) · International Conference on Language Resources and Evaluation · doi:10.48550/arxiv.2403.18121 · Cited by 19
“…Azaria, 2022; Qin et al., 2023; Wang et al., 2023; Lai et al., 2023; Tabone and De Winter, 2023) to interest in other areas of research such as regulation (Hacker et al., 2023), ethics (Zhuo et al., 2023), law (Choi et al., 2023), education (Susnjak, 2022), and medicine (Jeblick et al., 2022).”
BackgroundUtilizing Large Language Models in Tribal Emergency Management
Gupta, S., Chen, Y.-C., & Tsai, C.-H. (2024) · IUI Companion · doi:10.1145/3640544.3645219 · Cited by 19
Knowledge Bases and Language Models: Complementing Forces
Suchanek, F. M., & Tuan, L. A. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-45072-3_1 · Cited by 15
“…generated text [4], liability for generated statements [78], explainability of decisions based on such text, the personal data stored in language models (or submitted in prompts), the intellectual property of data used for training, and the remuneration of those who created the training data [20].”
BackgroundHuo, B., Marfo, N. Y. A., Sylla, P. A., Calabrese, E., Kumar, S. S., Slater, B. J., Walsh, D. S., & Vosburg, R. W. (2024) · Surgical Endoscopy · doi:10.1007/s00464-024-11155-5 · Cited by 20
“Patient data must be handled ethically and in compliance with privacy regulations [20, 21].”
The Battle of LLMs: A Comparative Study in Conversational QA Tasks
Rangapur, A., & Rangapur, A. (2024) · arXiv.org · doi:10.48550/arxiv.2405.18344 · Cited by 18
“As double-edged swords, these models should be monitored, according to (Hacker et al., 2023) and (Shen et al., 2023).”
BackgroundWe are Building Gods: AI as the Anthropomorphised Authority of the Past
Öhman, C. (2024) · Minds and Machines · doi:10.1007/s11023-024-09667-z · Cited by 28
Mustapha, K. B., Mustapha, K. B., Yap, E. H., & Abakr, Y. A. (2024) · Interactive Technology and Smart Education · doi:10.1108/itse-10-2023-0198 · Cited by 33
Generative AI in Finance: Risks and Potential Solutions
Remolina, N. (2023) · Social Science Research Network · doi:10.2139/ssrn.4628235 · Cited by 18
“In practice, the latest proposal of the EU AI Act treats all foundation models as high-risk applications, which imposes additional and onerous obligations on AI developers to mitigate hypothetical risks that may never materialize [20].”
AI and the transformation of accountability and discretion in urban governance
Goldsmith, S., & Yang, J. (2026) · Urban Governance · doi:10.1016/j.ugj.2026.02.005 · Cited by 17
“In this context, the concentration of innovation power among a few leading technology corporations further complicates governance and intensifies risks to regulation and oversight ( Hacker et al., 2023 ; Widder & Nafus, 2023 ; Narechania & Sitara-man, 2023 ).”
Mirtaheri, S. L., Movahedkor, N., Shahbazian, R., Pascucci, V., & Pugliese, A. (2025) · Future Generation Computer Systems · doi:10.1016/j.future.2025.108107 · Cited by 18
Hondjeu, A. R. M., Zhao, Z. Y., Newton, L., Ajenkar, A., Hladkowicz, E., Ladha, K. S., Wijeysundera, D. N., & McIsaac, D. I. (2025) · Canadian Journal of Anesthesia/Journal canadien d anesthésie · doi:10.1007/s12630-025-02980-w · Cited by 19
“61 Figure 4 summarizes key limitations of LLMs in perioperative medicine, including hallucinations, bias, transparency issues, security risks, and environmental concerns.”
Casino, F. (2025) · Technology in Society · doi:10.1016/j.techsoc.2025.102956 · Cited by 18
Normative conflicts and shallow AI alignment
Millière, R. (2025) · Philosophical Studies · doi:10.1007/s11098-025-02347-3 · Cited by 16
“In humans, failures typically occur when auxiliary factors such as time pressure or cognitive load interfere with the deliberative process; but AI systems are not subject to the same constraints.”
Self-Destructive Language Model
Wang, Y.-H., Zhu, R.-Y., & Wang, T. (2025) · arXiv.org · doi:10.48550/arxiv.2505.12186 · Cited by 17
“…solutions include regulating the parametric distance between fine-tuned and original models [38, 63, 54, 8], mixing alignment data with fine-tuning data [4, 64, 21], prompting to mitigate potential harmful behavior [34], and filtering harmful content from fine-tuning data [13, 24, 25].”
Explainable AI: The Quest for Transparency in Business and Beyond
DeSimone, H., & León, M. (2024) · International Conference on Information and Computer Technologies (ICICT) · doi:10.1109/icict62343.2024.00093 · Cited by 20
“Additionally, Hacker (2023) highlights the transformative prospects as well as risks associated with large generative AI models (LGAIMs), such as ChatGPT, and how current regulations are not suited to manage this class of AI [14].”
BackgroundEngaging the many-hands problem of generative-AI outputs: a framework for attributing credit
Khosrowi, D., Finn, F., & Clark, E. (2024) · AI and Ethics · doi:10.1007/s43681-024-00440-7 · Cited by 17
“…push the act’s regulatory obligations onto European deployers (e.g. app developers whose products access GenAI systems through APIs) and users of their general AI models (including the likes of GPT-4 and Stable Diffusion), rather than taking accountability for potential damages themselves [35, 70].”
Adiasto, K. (2024) · Group & Organization Management · doi:10.1177/10596011241238792 · Cited by 20
“As research conducted in the U.S. is readily generalizable to the rest of the world (yes, this is sarcasm), the rise of generative AI thus presents a compelling challenge for global work practices and policies (Dwivedi et al., 2023; Hacker et al., 2023).”
X-Mark: Towards Lossless Watermarking Through Lexical Redundancy
Chen, L., Bian, Y.-T., Deng, Y., Li, S., Wu, B.-Z., Zhao, P., & Wong, K.-F. (2023) · arXiv.org · doi:10.48550/arxiv.2311.09832 · Cited by 17
“Regulatory bodies can later extract this information to trace and regulate AI-generated content (Atal-lah et al., 2001; Wilson et al., 2014; Hacker et al., 2023).”
BackgroundAI providers as criminal essay mills? Large language models meet contract cheating law
Gaumann, N., & Veale, M. (2024) · Information & Communications Technology Law · doi:10.1080/13600834.2024.2352692 · Cited by 17
“155 They may not benefit from immunities, but also will not be subject to new proactive obligations, as other intermediaries have been.”
When, What, and how should generative artificial intelligence explain to Users?
Jang, S., Lee, H., Kim, Y., Lee, D., Shin, J., & Nam, J. (2024) · Telematics and Informatics · doi:10.1016/j.tele.2024.102175 · Cited by 20
Teaching UML using a RAG-based LLM
Ardimento, P., Bernardi, M., & Cimitile, M. (2024) · IEEE International Joint Conference on Neural Network · doi:10.1109/ijcnn60899.2024.10651492 · Cited by 16
Trustworthy Artificial Intelligence in the Context of Metrology
Adel, T., Bilson, S., Levene, M., & Thompson, A. (2024) · arXiv.org · doi:10.48550/arxiv.2406.10117 · Cited by 16
“It is therefore argued in [HEM23] that regulation should concentrate on the deployers and users of LLM technology rather than on the LLM developers themselves.”
BackgroundRU-AI: A Large Multimodal Dataset for Machine-Generated Content Detection
Huang, L., Zhang, Z., Zhang, Y., Zhou, X., & Wang, S. (2025) · International World Wide Web Conference (WWW) · doi:10.1145/3701716.3715306 · Cited by 14
“However, generative AI models’ capabilities may be misused for generating discriminate or disrespectful content, spreading misinformation, and conducting target scams.”
Sonkor, M. S., & de Soto, B. G. (2024) · International Journal of Construction Management · doi:10.1080/15623599.2024.2355782 · Cited by 15
“The authors also underlined that ChatGPT can potentially generate content harmful to society or individuals (Hacker et al. 2023; Lambert and Stevens 2023).”
Evolving Generative AI: Entangling the Accountability Relationship
Elliott, M. T. J., Deepak, P., & MacCárthaigh, M. (2024) · Digital Government Research and Practice · doi:10.1145/3664823 · Cited by 17
Prioritizing Software Requirements Using Large Language Models
Sami, M., Rasheed, Z., Waseem, M., Zhang, Z.-Y., Herda, T., & Abrahamsson, P. (2024) · arXiv.org · doi:10.48550/arxiv.2405.01564 · Cited by 16
“Generative AI, aims to produce new material by mimicking human-like outputs in a variety of fields, including computer vision, NLP, and the creation of images and videos [15].”
BackgroundFrom bias to black boxes: understanding and managing the risks of AI – an actuarial perspective
du Preez, V., Bennet, S., Byrne, M., Couloumy, A., Das, A., Dessain, J., Galbraith, R., King, P., Mutanga, V., Schiller, F., et al. (2024) · British Actuarial Journal · doi:10.1017/s1357321724000060 · Cited by 15
Chamola, V., Dave, D., Goyal, I., & Sharma, S. (2026) · Expert Systems · doi:10.1111/exsy.70201 · Cited by 28
Uses, opportunities and risks of artificial intelligence in participatory urban planning
Raymond, C. M., Nummi, P., von Wirth, T., Poom, A., Ahdekivi, A., Barthel, S., Delmelle, E., Dunkel, E., Fagerholm, N., Grêt-Regamey, A., et al. (2025) · Discover Cities · doi:10.1007/s44327-025-00137-4 · Cited by 15
“Collectively, these models allow AI to find patterns and relationships in data, generating new examples that are similar to the training data [40].”
Future Trends of Open-Source AI in Libraries: Implications for Librarianship and Service Delivery
Okwu, E., Oyighan, D., & Oladokun, B. (2024) · Asian Journal of Information Science and Technology · doi:10.70112/ajist-2024.14.2.4283 · Cited by 15
A Tool for Test Case Scenarios Generation Using Large Language Models
Sami, A. M., Rasheed, Z., Waseem, M., Zhang, Z.-Y., Tomas, H., & Abrahamsson, P. (2024) · arXiv.org · doi:10.48550/arxiv.2406.07021 · Cited by 15
“The transformer architecture, which greatly improves NLP skills by recognizing contextual relationships in text, is essential to the GPT model’s capacity to produce text that nearly resembles human writing Radford et al. [2018], Rasheed et al.. Recent advancements highlight how versatile and highly effective GPT model…”
BackgroundInfluentialExperimenting with Multi-Agent Software Development: Towards a Unified PlatformPreprint
Sami, M. A., Waseem, M., Rasheed, Z., Saari, M., Systä, K., & Abrahamsson, P. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2406.05381 · Cited by 20
“Generative AI assists in creating outputs that mimic human performance across diverse fields including multimedia production, computer vision, and NLP [14].”
BackgroundWhat is it for a Machine Learning Model to Have a Capability?
Harding, J., & Sharadin, N. (2024) · British Journal for the Philosophy of Science · doi:10.48550/arxiv.2405.08989 · Cited by 15
“For readability’s sake, we drop the ‘general-purpose’ prefix in what follows.”
Rajan, B., Carradını, S., & Lauer, C. (2024) · CHI Conference Extended Abstracts on Human Factors in Computing Systems (CHI EA) · doi:10.1145/3613905.3650919 · Cited by 15
“Upon receiving a user query, the bot goes through three meticulous checks that serve as guardrails to ensure the safety and integrity of the responses [16].”
Method
Metadata
- Authors
- Hacker, P., Engel, A., & Mauer, M.
- Affiliations
- European University Viadrina (DE)
- Heidelberg University (DE)
- Venue
- 2023 ACM Conference on Fairness Accountability and Transparency
- Publisher
- ACM
- Published
- 2023-06-12
- Pages
- 1112-1123
- OpenAlex
- W4380319827
- Cited by
- 466
- References
- 176
- Citations by year
- 2026: 59 · 2025: 151 · 2024: 191 · 2023: 65
- FWCI
- 1289.86
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- ACM Conference on Fairness, Accountability, and Transparency (FAccT) (publishedVersion, cc-by-nc-sa)
- arXiv (Cornell University) (submittedVersion, cc-by-nc-sa)
- arXiv (Cornell University)
- Funders
- Universität Hamburg
- Topics
- Social Sciences › Law › Digitalization, Law, and Regulation
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Physical Sciences › Computer Science › Artificial Intelligence › Law, AI, and Intellectual Property
- Keywords
- ChatGPT, AI regulation, trustworthy AI, EU AI Act, AI value chain, high-risk AI systems, data protection, content moderation, Digital Services Act, transparency obligations, risk management, non-discrimination, pre-trained models
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
