CoChat
Sign Up
Cited by 124Digitalization, Law, and RegulationLaw, AI, and Intellectual PropertyEthics and Social Impacts of AI

Generative AI in EU law: Liability, privacy, intellectual property, and cybersecurity

Open access
Novelli, Claudio, Casolari, Federico, Hacker, Philipp, Spedicato, Giorgio, Floridi, Luciano·2024·Computer Law & Security Review
DOI 10.1016/j.clsr.2024.106066

Published version of a preprint

Open this paper in CoChat — free account

Abstract

The complexity and emergent autonomy of Generative AI systems introduce challenges in predictability and legal compliance. This paper analyses some of the legal and regulatory implications of such challenges in the European Union context, focusing on four areas: liability, privacy, intellectual property, and cybersecurity. It examines the adequacy of the existing and proposed EU legislation, including the Artificial Intelligence Act (AIA), in addressing the challenges posed by Generative AI in general and LLMs in particular. The paper identifies potential gaps and shortcomings in the EU legislative framework and proposes recommendations to ensure the safe and compliant deployment of generative models.

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.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1016/j.clsr.2024.106066

Cites

Statistics

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.

A few free checks a day · papers already verified open instantly · results are public

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 revised PLD treats AI software as a product and introduces disclosure, burden-shifting, and transparency obligations (Articles 9-13), helping victims establish liability in cases of AI-related damage.

    “The revised Directive treats AI software as a “product” and introduces disclosure, burden-shifting, and transparency obligations (Articles 9-13), helping victims establish liability in cases of AI-related damage (Novelli et al., 2024).”

    From Deepfakes, Democracy, and the Ethics of Synthetic Media: A Synthesis of the SOLARIS Project (2026) · cites this paper as Novelli et al., 2024

    Section 2 (p. 2) says the proposed PLD extends to AI systems and AI-enabled goods and describes disclosure, rebuttable presumptions, and burden-shifting that can help claimants. But it excludes some open-source software and locates PLD disclosure under Article 8, burden-of-proof rules under Article 9, and transparency under Articles 9 and 13—not the claimed Articles 9–13 as a group. The paper also describes the revised PLD as a proposal with an agreed text, rather than an operative directive. (LLM, conf=0.86)

    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.

No citation lists for this paper

No citation lists for this paper yet — checked again daily.

References 86 · Cited by 124

Research works printed but not linked: 73

  1. [1][1] [Foster D. Generative deep learning: teaching machines to paint, write, compose, and play. Sebastopol, CA: O ' Reilly Media; 2019.](http://refhub.elsevier.com/S0267-3649(24)00132-8/sbref0001) -
  2. [2][2] Hacker P, Engel A, Mauer M. Regulating ChatGPT and other large generative AI models. In: Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency, in FAccT ' 23. Association for Computing Machinery; Jun. 2023. p. 1112 -23. https://doi.org/10.1145/3593013.3594067. - · DOI · Paper page
  3. [4][4] O. Dheu, J. De Bruyne, and C. Ducuing, 'The european commission ' s approach to extra-contractual liability and AI -A first analysis and evaluation of the two proposals ' , Oct. 06, 2022, Rochester, NY: 4239792. doi:10.2139/ssrn.4239792. - · DOI
  4. [5][5] Novelli C, Casolari F, Rotolo A, Taddeo M, Floridi L. AI risk assessment: a scenariobased, proportional methodology for the AI act. Digit. Soc. Mar. 2024;3(1):13. https://doi.org/10.1007/s44206-024-00095-1. - · DOI
  5. [6][6] Novelli C, Casolari F, Rotolo A, Taddeo M, Floridi L. Taking AI risks seriously: a new assessment model for the AI Act. AI Soc. Jul. 2023. https://doi.org/10.1007/ s00146-023-01723-z. -
  6. [9][9] C. Novelli, F. Casolari, A. Rotolo, M. Taddeo, and L. Floridi, 'Taking AI Risks seriously: a proposal for the AI act ' , May 14, 2023, Rochester, NY: 4447964. doi:1 0.2139/ssrn.4447964. -
  7. [11][11] C. Lee, K. Cho, and W. Kang, 'Mixout: effective regularization to finetune largescale pretrained language models ' , Jan. 22, 2020, arXiv: arXiv:1909.11299. doi:1 0.48550/arXiv.1909.11299. - · arXiv
  8. [12][12] Y. Xiao, P.P. Liang, U. Bhatt, W. Neiswanger, R. Salakhutdinov, and L.-P. Morency, 'Uncertainty quantification with pre-trained language models: a large-scale empirical analysis ' , Oct. 14, 2022, arXiv: arXiv:2210.04714. doi:10.4855 0/arXiv.2210.04714. - · arXiv
  9. [13][13] Hacker P. What ' s missing from the EU AI act: addressing the four key challenges of large language models. Verfassungsblog Dec. 2023. https://doi.org/10.17176/ 20231214-111133-0. -
  10. [14][14] Falco G, et al. Governing AI safety through independent audits. Nat Mach Intell 2021;3(7). https://doi.org/10.1038/s42256-021-00370-7. Art. no. 7, Jul. - · DOI · Paper page
  11. [15][15] Novelli C, Taddeo M, Floridi L. Accountability in artificial intelligence: what it is and how it works. AI Soc Feb. 2023. https://doi.org/10.1007/s00146-023-01635y. - · DOI
  12. [17][17] T.B. Brown et al., 'Language models are few-shot learners ' , Jul. 22, 2020, arXiv: arXiv:2005.14165. doi:10.48550/arXiv.2005.14165. - · DOI · Paper page
  13. [19][19] Brown H, Lee K, Mireshghallah F, Shokri R, Tram ` er F. What does it mean for a language model to preserve privacy?. In: 2022 ACM Conference on Fairness, Accountability, and Transparency, in FAccT ' 22. Association for Computing Machinery; Jun. 2022. p. 2280 -92. https://doi.org/10.1145/3531146.3534642. - · DOI · Paper page
  14. [20][20] Carlini N, et al. Extracting training data from diffusion models. In: Presented at the 32nd USENIX Security Symposium (USENIX Security 23); 2023. p. 5253 -70. Accessed: Jan. 13, 2024. [Online]. Available: https://www.usenix.org/conference /usenixsecurity23/presentation/carlini. -
  15. [21][21] Feldman RC, Hyman DA, Price WN, Ratain MJ. Negative innovation: when patents are bad for patients. Nat Biotechnol 2021;39(8). https://doi.org/10.1038/s41587021-00999-0. Art. no. 8, Aug. - · DOI
  16. [22][22] Fredrikson M, Jha S, Ristenpart T. Model inversion attacks that exploit confidence information and basic countermeasures. In: Proceedings of the 22nd ACM SIGSAC Conference on Computer and Communications Security, in CCS ' 15. Association for Computing Machinery; Oct. 2015. p. 1322 -33. https://doi.org/10.1145/ 2810103.2813677. -
  17. [23][23] Oostveen M. Identifiability and the applicability of data protection to big data. Int Data Priv Law Nov. 2016;6(4):299 -309. https://doi.org/10.1093/idpl/ipw012. - · DOI
  18. [25][25] F. Zuiderveen Borgesius, S. Kruikemeier, S. Boerman, and N. Helberger, 'Tracking walls, take-it-or-leave-it choices, the GDPR, and the ePrivacy regulation ' , Mar. 15, 2018, Rochester, NY: 3141290. Accessed: Jan. 13, 2024. [Online]. Available: htt ps://papers.ssrn.com/abstract = 3141290. -
  19. [27][27] Peloquin D, DiMaio M, Bierer B, Barnes M. Disruptive and avoidable: GDPR challenges to secondary research uses of data. Eur J Hum Genet 2020;28(6). https://doi.org/10.1038/s41431-020-0596-x. Art. no. 6, Jun. - · DOI
  20. [28][28] Donnelly M, McDonagh M. Health research, consent and the GDPR exemption. Eur J Health Law Apr. 2019;26(2):97 -119. https://doi.org/10.1163/1571809312262427. - · DOI
  21. [29][29] Hacker P. A legal framework for AI training data -From first principles to the artificial intelligence act. Law Innov Technol Jul. 2021;13(2):257 -301. https://doi. org/10.1080/17579961.2021.1977219. - · DOI
  22. [30][30] T. Zarsky, 'Incompatible: the GDPR in the age of big data ' , Aug. 08, 2017, Rochester, NY: 3022646. Accessed: Jan. 13, 2024. [Online]. Available: https:// papers.ssrn.com/abstract = 3022646. -
  23. [31][31] Bonatti PA, Kirrane S. Big data and analytics in the age of the GDPR. In: 2019 IEEE Int. Congr. Big Data BigDataCongress; Jul. 2019. p. 7 -16. https://doi.org/ 10.1109/BigDataCongress.2019.00015. - · DOI
  24. [32][32] [Bederman DJ. The souls of international organizations: legal personality and the lighthouse at cape spartel. International legal personality. Routledge; 2010.](http://refhub.elsevier.com/S0267-3649(24)00132-8/sbref0032) -
  25. [33][33] E. Lehman, S. Jain, K. Pichotta, Y. Goldberg, and B.C. Wallace, 'Does BERT pretrained on clinical notes reveal sensitive data? ' , Apr. 22, 2021, arXiv: arXiv:2 104.07762. doi:10.48550/arXiv.2104.07762. - · DOI
  26. [35][35] Bi B, Shokouhi M, Kosinski M, Graepel T. Inferring the demographics of search users: social data meets search queries. In: Proceedings of the 22nd international conference on World Wide Web, in WWW ' 13. Association for Computing Machinery; May 2013. p. 131 -40. https://doi.org/10.1145/2488388.2488401. - · DOI
  27. [36][36] Chaturvedi R, Chaturvedi S. It ' s all in the name: a character-based approach to infer religion. Polit Anal Jan. 2024;32(1):34 -49. https://doi.org/10.1017/ pan.2023.6. -
  28. [38][38] M. Zhang, O. Press, W. Merrill, A. Liu, and N.A. Smith, 'How language model hallucinations can snowball ' , May 22, 2023, arXiv: arXiv:2305.13534. doi:10.4855 0/arXiv.2305.13534. - · arXiv
  29. [39][39] Villaronga EF, Kieseberg P, Li T. Humans forget, machines remember: artificial intelligence and the right to be forgotten. Comput Law Secur Rev Apr. 2018;34(2): 304 -13. https://doi.org/10.1016/j.clsr.2017.08.007. - · DOI
  30. [40][40] E. De Cristofaro, 'An overview of privacy in machine learning ' , May 18, 2020, arXiv: arXiv:2005.08679. doi:10.48550/arXiv.2005.08679. - · DOI
  31. [41][41] E. Hine, C. Novelli, M. Taddeo, and L. Floridi, 'Supporting trustworthy AI through machine unlearning ' , Nov. 24, 2023, Rochester, NY: 4643518. doi:10.2139/ssr n.4643518. - · DOI
  32. [43][43] Floridi L. Machine unlearning: its nature, scope, and importance for a ' delete culture. Philos Technol Jun. 2023;36(2):42. https://doi.org/10.1007/s13347-02300644-5. - · DOI
  33. [44][44] [Malgieri G. Vulnerability and Data Protection Law. Oxford data protection & privacy law. Oxford, New York: Oxford University Press; 2023.](http://refhub.elsevier.com/S0267-3649(24)00132-8/sbref0044) -
  34. [45][45] K. de la Durantaye, ' ' Garbage In, Garbage Out ' - Die regulierung generativer KI durch Urheberrecht ' , Aug. 10, 2023, Rochester, NY: 4571908. Accessed: Jan. 13, 2024. [Online]. Available: https://papers.ssrn.com/abstract = 4571908. -
  35. [46][46] G. Sartor, F. Lagioia, and G. Contissa, 'The use of copyrighted works by AI systems: art works in the data mill ' , Oct. 11, 2018, Rochester, NY: 3264742. doi:10.21 39/ssrn.3264742. -
  36. [48][48] T. Klawonn, 'Urheberrechtliche Grenzen des Web Scrapings (Web Scraping under German Copyright Law) ' , Nov. 20, 2019, Rochester, NY: 3491192. doi:10.2139/ssr n.3491192. - · DOI
  37. [49][49] Dermawan A. Text and data mining exceptions in the development of generative AI models: what the EU member states could learn from the Japanese ' nonenjoyment ' purposes? J World Intellect Prop 2024;27(1):44 -68. https://doi.org/10.1111/ jwip.12285. -
  38. [50][50] [Pesch PJ, B ¨ ohme R. Artpocalypse now? - Generative KI und die vervielf ¨ altigung von trainingsbildern. Gewerbl Rechtsschutz Urheberrecht GRUR 2023:997 -1007.](http://refhub.elsevier.com/S0267-3649(24)00132-8/sbref0050) -
  39. [51][51] C. Geiger, G. Frosio, and O. Bulayenko, 'The exception for text and data mining (TDM) in the proposed directive on copyright in the digital single market - legal aspects ' , Mar. 02, 2018, Rochester, NY: 3160586. doi:10.2139/ssrn.3160586. - · DOI
  40. [52][52] Rosati E. Article 4 -Exception or limitation for text and data mining. In: Rosati E, editor. Copyright in the digital single market: article-by-article commentary to the provisions of directive 2019/790. Oxford University Press; 2021. https://doi.org/ 10.1093/oso/9780198858591.003.0005. editorp. 0. - · DOI
  41. [53][53] J. Griffiths, 'The ' Three-Step Test ' in european copyright law - problems and solutions ' , Sep. 22, 2009, Rochester, NY: 1476968. Accessed: Jan. 13, 2024. [Online]. Available: https://papers.ssrn.com/abstract = 1476968. -
  42. [54][54] Marcus G, Southen R. Generative AI has a visual plagiarism problem. IEEE Spectr Jan. 06, 2024. Accessed: Jan. 13, 2024. [Online]. Available: https://spectrum.ieee. org/midjourney-copyright. -
  43. [55][55] J. Gillotte, 'Copyright infringement in AI-generated artworks ' , Jun. 01, 2020, Rochester, NY: 3657423. Accessed: Aug. 13, 2023. [Online]. Available: https:// papers.ssrn.com/abstract = 3657423. -
  44. [56][56] P. Henderson, X. Li, D. Jurafsky, T. Hashimoto, M.A. Lemley, and P. Liang, 'Foundation models and fair use ' , Mar. 27, 2023, arXiv: arXiv:2303.15715. doi:10.48550/arXiv.2303.15715. - · DOI
  45. [58][58] [Nordemann JB. Neu: t ¨ aterschaftliche haftung von hostprovidern im urheberrecht bei (Verkehrs-)Pflichtverletzungen im internet. ZUM; 2023.](http://refhub.elsevier.com/S0267-3649(24)00132-8/sbref0058) -
  46. [59][59] Engel A. Can a patent be granted for an AI-generated invention? GRUR Int Nov. 2020;69(11):1123 -9. https://doi.org/10.1093/grurint/ikaa117. - · DOI
  47. [60][60] [Hristov K. Artificial intelligence and the copyright dilemma. IDEA J Frankl Pierce Cent Intellect Prop 2016;57(3):431 -54. 2017.](http://refhub.elsevier.com/S0267-3649(24)00132-8/sbref0060) -
  48. [61][61] Varytimidou C. The new A(I)rt movement and its copyright protection: immoral or E-moral? GRUR Int Mar. 2023;72(4):357 -63. https://doi.org/10.1093/grurint/ ikac153. - · DOI
  49. [62][62] Hugenholtz PB, Quintais JP. Copyright and artificial creation: does EU copyright law protect AI-assisted output? IIC - Int Rev Intellect Prop Compet Law Oct. 2021; 52(9):1190 -216. https://doi.org/10.1007/s40319-021-01115-0. - · DOI
  50. [63][63] A. Ramalho, 'Patentability of AI-generated inventions: is a reform of the patent system needed? ' , Feb. 15, 2018, Rochester, NY. doi:10.2139/ssrn.3168703. - · DOI
  51. [65][65] T.W. Dornis, 'Of ' Authorless Works ' and ' Inventions without Inventor ' -The muddy waters of ' AI autonomy ' in intellectual property doctrine ' , Jan. 29, 2021, Rochester, NY: 3776236. doi:10.2139/ssrn.3776236. - · DOI
  52. [66][66] Hilty RM, Hoffmann J, Scheuerer S. Intellectual property justification for artificial intelligence. In: Lee J-A, Hilty R, Liu K-C, editors. Artificial intelligence and intellectual property. Oxford University Press; 2021. https://doi.org/10.1093/oso/ 9780198870944.003.0004. p. 0. - · DOI
  53. [67][67] Hoffmann-Riem W. Artificial intelligence as a challenge for law and regulation. In: Wischmeyer T, Rademacher T, editors. Regulating Artificial Intelligence. Cham: Springer International Publishing; 2020. p. 1 -29. https://doi.org/10.1007/978-3030-32361-5\_1. - · DOI
  54. [68][68] P.R. Goold, 'The curious case of computer-generated works under the copyright, designs and patents act 1988 ′ , Jan. 01, 2021, Rochester, NY: 4072004. Accessed: Aug. 13, 2023. [Online]. Available: https://papers.ssrn.com/abstract = 4072004. -
  55. [69][69] Guadamuz A. 'Do androids dream of electric copyright? Comparative analysis of originality in artificial intelligence generated works. In: Lee J-A, Hilty R, Liu K-C, editors. Artificial intelligence and intellectual property. Oxford University Press; 2021. https://doi.org/10.1093/oso/9780198870944.003.0008. p. 0. - · DOI · Paper page
  56. [70][70] [Spedicato G. L ' attivit ` a di web scraping nelle banche dati ed il riuso delle informazioni. Riv Dirit Ind 2019;4 -5:253 -307.](http://refhub.elsevier.com/S0267-3649(24)00132-8/sbref0070) -
  57. [71][71] Purtova N. The law of everything. Broad concept of personal data and future of EU data protection law. Law Innov Technol Jan. 2018;10(1):40 -81. https://doi.org/ 10.1080/17579961.2018.1452176. - · DOI
  58. [72][72] Joint Research Centre (European Commission) etal., Cybersecurity of artificial intelligence in the AI Act: guiding principles to address the cybersecurity requirement for high risk AI systems. LU: publications Office of the European Union, 2023. Accessed: Jan. 13, 2024. [Online]. Available: https://data.europa.eu /doi/10.2760/271009. - · DOI
  59. [73][73] E. Wallace, S. Feng, N. Kandpal, M. Gardner, and S. Singh, 'Universal adversarial triggers for attacking and analyzing NLP ' , Jan. 03, 2021, arXiv: arXiv:1908.07125. doi:10.48550/arXiv.1908.07125. - · DOI
  60. [74][74] J.A. Goldstein, G. Sastry, M. Musser, R. DiResta, M. Gentzel, and K. Sedova, 'Generative language models and automated influence operations: emerging threats and potential mitigations ' , Jan. 10, 2023, arXiv: arXiv:2301.04246. doi:10.48550/arXiv.2301.04246. - · DOI · Paper page
  61. [75][75] L. Weidinger et al., 'Ethical and social risks of harm from language models ' , Dec. 08, 2021, arXiv: arXiv:2112.04359. doi:10.48550/arXiv.2112.04359. - · DOI · Paper page
  62. [76][76] P. Ranade, A. Piplai, S. Mittal, A. Joshi, and T. Finin, 'Generating fake cyber threat intelligence using transformer-based models ' , Jun. 18, 2021, arXiv: arXiv:2102 .04351. doi:10.48550/arXiv.2102.04351. - · DOI
  63. [79][79] V. Adlakha, P. BehnamGhader, X.H. Lu, N. Meade, and S. Reddy, 'Evaluating correctness and faithfulness of instruction-following models for question answering ' , Jul. 31, 2023, arXiv: arXiv:2307.16877. doi:10.48550/arXiv.2 307.16877. - · DOI
  64. [81][81] S.M.T.I. Tonmoy et al., 'A comprehensive survey of hallucination mitigation techniques in large language models ' , Jan. 03, 2024, arXiv: arXiv:2401.01313. doi:10.48550/arXiv.2401.01313. - · DOI · Paper page
  65. [82][82] X. Li et al., 'Self-alignment with instruction backtranslation ' , Aug. 14, 2023, arXiv: arXiv:2308.06259. doi:10.48550/arXiv.2308.06259. - · DOI
  66. [84][84] H. Kang, J. Ni, and H. Yao, 'Ever: mitigating hallucination in large language models through real-time verification and rectification ' , Nov. 15, 2023, arXiv: arXiv:2 311.09114. doi:10.48550/arXiv.2311.09114. - · DOI
  67. [86][86] R. Cohen, M. Hamri, M. Geva, and A. Globerson, 'LM vs LM: detecting Factual Errors via Cross Examination ' , May 22, 2023, arXiv: arXiv:2305.13281. doi:10.48550/arXiv.2305.13281. - · DOI · Paper page
  68. [87][87] Eckhardt P, Kotovskaia A. The EU ' s cybersecurity framework: the interplay between the cyber resilience act and the NIS 2 Directive. Int Cybersec Law Rev Jun. 2023;4(2):147 -64. https://doi.org/10.1365/s43439-023-00084-z. - · DOI
  69. [88][88] C. Novelli, P. Hacker, J. Morley, J. Trondal, and L. Floridi, 'A robust governance for the AI act: AI office, AI Board, scientific panel, and national authorities ' , May 05, 2024, Rochester, NY: 4817755. doi:10.2139/ssrn.4817755. - · DOI
  70. [89][89] [Leboukh F, Aduku E, Ali O. Balancing ChatGPT and data protection in Germany: challenges and opportunities for policy makers. J Politics Ethics New Technol AI 2023;2(1). e35166-e35166.](http://refhub.elsevier.com/S0267-3649(24)00132-8/sbref0089) -
  71. [90][90] Hacker P, Borgesius FZ, Mittelstadt B, Wachter S. Generative discrimination: what happens when generative AI exhibits bias, and what can be done about it, forthcoming. In: Hacker Philipp, Engel Andreas, Hammer Sarah, Mittelstadt Brent, editors. The Oxford handbook of the foundation and regulation of generative AI. Oxford University Press; 2024. forthcoming, https://papers.ssrn.com/sol3/papers. cfm?abstract\_id = 4877398. Part IV.2. -
  72. [92][92] [Paulina JP, Rainer B. Verarbeitung personenbezogener Daten und datenrichtigkeit bei gro ß en sprachmodellen. MMR 2023:917.](http://refhub.elsevier.com/S0267-3649(24)00132-8/sbref0092) -
  73. [93][93] [Moos Flemming. Personenbezug von large language models. CR 2024:442.](http://refhub.elsevier.com/S0267-3649(24)00132-8/sbref0093)

Authors
Novelli, C., Casolari, F., Hacker, P., Spedicato, G., & Floridi, L.
Affiliations
  • Yale University (US)
  • University of Bologna (IT)
  • European University Viadrina (DE)
Corresponding author
Novelli, Claudio
ORCID
Venue
Computer Law & Security Review
Publisher
Elsevier BV
Published
2024-11
Volume
55
Pages
106066
ISSN
2212-473X (ISSN-L)
DOI
10.1016/j.clsr.2024.106066
OpenAlex
W4403714226
Cited by
124
References
86
Citations by year
2026: 62 · 2025: 50 · 2024: 12
FWCI
317.52
Type
journal article
Language
en
Open access
Licence
Locations
  • Computer law & security review (publishedVersion, cc-by)
  • SSRN Electronic Journal (acceptedVersion)
  • Archivio istituzionale della ricerca (Alma Mater Studiorum Università di Bologna) (submittedVersion, cc-by-nc-nd)
Topics
  • Social Sciences › Law › Digitalization, Law, and Regulation
  • Physical Sciences › Computer Science › Artificial Intelligence › Law, AI, and Intellectual Property
  • Social Sciences › Safety Research › Ethics and Social Impacts of AI
Keywords
generative AI, large language models, European Union law, liability, privacy, intellectual property, cybersecurity, EU AI Act, AI regulation, legal compliance, legislative gaps

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