CoChat

Artificial Infringement

Open access
Goodyear, Michael · 2025 · SSRN Electronic Journal

The DOI matches its Crossref record; 51 of the 55 references it cites were found in a registry or an index.

Abstract

Generative AI is changing the way we do everything from legal research to artistic creation. This is possible through recent advances in machine learning that allow AI systems to program themselves. With greater AI capacity, however, comes increasingly unpredictable outputs. AI systems will often generate an output the user and the developer never considered. Sometimes, these unforeseen outputs can infringe others’ copyrights in creative works. In the past two years, copyright law has become one of the leading legal and policy battlegrounds for generative AI. Yet the question of who should be liable when AI systems infringe has barely been addressed. By examining the historical and doctrinal response of copyright law to new technologies, this Article offers a new analytical framework for determining liability for what it terms artificial infringement, or infringing outputs created by generative AI systems. Time and again, new technologies have posed challenges to existing copyright law, straining its capacity to both protect authors’ rights to incentivize new creative works and provide public access to those works. Courts and Congress have been able to maintain this balance by using a variety of doctrinal tools, including fair use, compulsory licensing, and secondary liability. One undertheorized tool, however, is the refinement of the copyright infringement claim. Courts introduced the “volition or causation” requirement to balance copyright in response to the rise of complex machine-generated infringements. This Article proposes that the AI system should be held directly liable for artificial infringement because it caused the infringing expression to occur. By making the AI system the direct infringer, courts can remove copyright law from strict liability and instead utilize and refine secondary liability doctrines to conduct a more nuanced, fault-based analysis of user and developer liability for AI-generated infringements. Together with the fair use doctrine, this conceptualization of the AI system as the direct infringer and users and developers as potentially secondarily liable provides a more comprehensive resolution to the existential infringement battles between copyright owners and AI while maintaining a balance between copyright’s competing policy goals.

Overview

Integrity signals

Registry

Crossref · 10.2139/ssrn.5184405

Cites

51 of 51 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Verification

Claims the paper makes

We couldn’t retrieve this paper’s full text, so its claims and self-check weren’t run. The registry and citation checks still apply.

Citations

Cites

Research works printed but not linked: 103

  1. Casey Lemley, 1324, Rachum-Twaig, supra note 34, at 1146-47. 48 Will Knight, The Dark Secret at the Heart of AI, MIT TECH. REV, L. REV, 83
  2. Surden, Calo, supra note 43
  3. Gary Marcus, 2018, What Is Deep Learning?, #:~:text=Deep%20learning%20is%20a%20method
  4. Calo, 2021, Support of, 1
  5. M Jack, 2015, The Path of Robotics Law, 6 CALIF, L. REV. CIR, 45
  6. Ziv Epstein, 1110, & Investigators of Human Creativity, Art and the Science of Generative AI, 380 SCI. 1110, 1110
  7. Hertzmann Epstein, Investigators of Human Creativity, supra note 61
  8. Mengchun Zhang, 2023, A Survey on Graph Diffusion Models: Generative AI in Science for Molecule, Protein and Material
  9. Jaron Lanier, 2023, There Is No A.I., NEW YORKER
  10. Emily M Bender, 2021, On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?, FACCT '21: PROCEEDINGS OF THE 2021 ACM CONFERENCE ON FAIRNESS, ACCOUNTABILITY, AND TRANSPARENCY, 610, 616
  11. Kevin Roose, 2022, Generated Picture Won an Art Prize. Artists Aren't Happy, TIMES
  12. Debra Cassens, 2023, Latest Version of ChatGPT Aces Bar Exam with Score Nearing 90th Percentile, ABA J
  13. Steve Brachmann, 2014, The Relevance of Algorithms, Senate IP Subcommittee Mulls Federal Right of Publicity at AI and Copyright Hearing, IPWATCHDOG, 167
  14. A Karni, 2019, Am I an Algorithm or a Product, When Products Liability Should Apply to Algorithmic Decision-Makers, 30 STAN. L. & POL'Y REV, 61
  15. Eric Avidon, 2023, Generative AI Hype Evolving Into Reality in Data
  16. Lanier, supra note 67; Grimmelmann, supra note 4
  17. Calo, 1331-33 (describing how traditionally strict liability requires foreseeability yet AI injuries are often unforeseeable), 554
  18. Mindy Nunez, 2023, Malpractice by the Autonomous AI Physician, J.L., TECH. & POL'Y 1
  19. Bryce Goodman, 2016, European Union Regulations on Algorithmic Decision-Making and a "Right to Explanation
  20. Abraham & Rabin, supra note 12, at 147-60 (proposing a manufacturerfinanced strict liability compensation regime referred to as, Manufacturer Enterprise Responsibility
  21. C David, 2014, Machines Without Principals: Liability Rules and Artificial Intelligence, 89
  22. Yale J Health Pol'y, 2021, proposing a new FDA regulatory framework for AI-driven medical devices, including more stringent pre-market approvals, black box-specific requirements, and monitoring and improvement obligations, & ETHICS, 433
  23. Duffourc, at 20 (proposing a control-centered framework and limited legal personhood for Autonomous AI Physicians), 143
  24. E G See, 2010, Technological Fair Use, 83 S. CAL. L. REV, 797, 800
  25. Jacob Victor, 1900, ) (emphasizing the importance of fair use for permitting utilityexpanding technologies such as search engines, plagiarism checkers, Utility-Expanding Fair Use, 105 MINN. L. REV. 1887
  26. Pamela Samuelson, 2009, Unbundling Fair Uses, 77 FORDHAM L. REV. 2537
  27. Edward Lee, 2018, Fair Use Avoidance in Music Cases, 59 B.C. L. REV, 171 NIMMER & NIMMER, 1873
  28. 172 Netcom, 907 F. Supp. at 1365. 173 Id, 1365
  29. 2019, concurring) (referring to Netcom and concluding that "volition and causation are different concepts, 52
  30. E D Pa, 2006, aff'd 242 F, 833
  31. Ltd Boisson V. Banian, 2001, ) (explaining how access and "similarities that are probative of copying, can show copyright infringement, 267
  32. 2014, (noting that causation "generally bars suits for alleged harm that is 'too remote' from the defendant's unlawful conduct, Lexmark Int'l, Inc. v. Static Control Components, 572 U.S, 118
  33. N D, 1995, Inc., 907 F. Supp. 1361, 1370
  34. Steven Shavell, 2009, ECONOMIC ANALYSIS OF ACCIDENT LAW, 297
  35. C Thomas, The Legitimate Function of the Public Tort
  36. Accidents, 1970, proposing the idea that an individual should internalize all potential costs of an action
  37. Restatement, THIRD) OF TORTS: PHYSICAL & EMOTIONAL HARM � 6 cmt
  38. Marchisio, The court does not find workable a theory of infringement that would hold the entire Internet liable for activities that cannot reasonably be deterred, Ctr. v. Netcom On-Line Comm'n Servs
  39. See Banks, 1985, Unless the factors of rational expectations, fault-whether legal or moral-and normative goals are to be disregarded in determining civil liability, foreseeability must remain a central inquiry, L. REV, 286
  40. See Mark, 2016, Hidden in Plain Sight: The Normative Source of Modern Tort Law, 91 N.Y.U. L. REV. 1517, 1553
  41. Restatement, THIRD) OF TORTS: PRODUCTS LIABILITY � 2(b)-(c)
  42. Balganesh, 228 In re Boeing 737 MAX Pilots Litig., 638 F
  43. Balganesh, 1984, 230 See supra Part II, 231
  44. C P John, 2020, RECOGNIZING WRONGS, 3
  45. C P John, 2006, Two Conceptions of Tort Damages: Fair v. Full Compensation, 55 DEPAUL L. REV, 435
  46. Guido Calabresi, 2022, On Tort Law's Dualisms, 135 HARV, L. REV. F, 184
  47. Under general tort-law principles, a person who orders or induces an actor's tortious conduct is subject to liability for harm resulting to a third person when the person 'knows or should know of circumstances that would make the conduct tortious if it were his own
  48. E G See, Artificial Stupidity, 61 WM. & MARY L. REV. 1187, 1193 (2020) (advocating for understanding optimal innovation policies in intellectual property to allow true artificial intelligence, rather than mere artificial stupidity, to be developed)
  49. E G See, 2018, 246 For more on the eight stages in the generative-AI supply chain and the various parties involved, DOMINGOS, supra note 40, 1, 28
  50. Abbott, 2016, 23; Matthew U. Scherer, Regulating Artificial Intelligence Systems: Risks, Challenges, Competencies, and Strategies, 354, 363
  51. E G See, 1106, ) (the element of access may be established through "widespread dissemination and subconscious copying" 256 Rosati
  52. Sag, 330-31; see also Guadamuz, supra note 44, 125
  53. Adi Robertson, 2022, Professional AI Whisperers Have Launched a Marketplace for DALL-E Prompts
  54. See Abraham & Rabin, supra note 12, at 130-31 (discussing the six tiered levels of automation adopted by the National Highway Traffic Safety Administration
  55. Davola, 2014, 264 See supra Part II.B. 265 Am, 599; Abraham & Rabin, supra note 12, 145
  56. Capitol Recs, 2013, LLC v. ReDigi Inc., 934 F. Supp. 2d 640
  57. Fromer Chatterjee, This trend and phenomenon is coupled with the unpredictable manner in which such robots behave and the inability to foresee risks that they may inflict, 286
  58. See Jason, 2018, (describing legal personhood as "a pre-existing legal fiction used to hold entities, not just natural persons, accountable"). Quasi-personhood also avoids the risk of considering an AI system property that could potentially be subject to an in rem action, J. HEALTH L. & ETHICS, 51
  59. 2009, Petroleos Mexicanos Refinacion v. M/T KING A, 554 F.3d 99, 108
  60. See Elizabeth, Corporate Personhood and Limited Sovereignty, 74
  61. B Paul, 2021, Corporate Personality, Purpose, and Liability, in RESEARCH HANDBOOK ON CORPORATE PURPOSE AND PERSONHOOD 222, 225
  62. Wyo. Stat. �, 17, 31
  63. Lawrence B Solum, 1992, Report with Recommendations to the Commission on Civil Law Rules on Robotics, 0051, � � 59(f), 294, 1499
  64. ; Balkin, 2023, Artificial Intelligence and Interspecific Law, 382 SCI, 376
  65. 1995, Religious Tech. Ctr. v. Netcom On-Line Comm'n Servs., Inc
  66. Katherine B Forrest, 2024, The Ethics and Challenges of Legal Personhood for AI, 133 YALE L, J. FORUM, 1175
  67. 2023, An individual may only obtain a narrow copyright in their modification of AI generated work after the fact or by using it as a tool they control. Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence, 16190
  68. Copyrightability Report, 18
  69. Thaler, 2023, WL, 5333236
  70. Gervais, 1995, 308 See supra Parts I and II. 309 See supra Part I.C. 310 Religious Tech, Ctr. v. Netcom On-Line Comm'n Servs., Inc., 907 F. Supp, 1361, 1372
  71. at 21 (expressing concern that treating AI as a product for purposes of tort liability would discourage automation). 313 17 U.S.C. � 504(c) (statutory damages of $750-$30,000, up to $150,000 for willful infringement
  72. See Anat Lior, 2022, The Role of Insurance in Artificial Intelligence Regulation, 467, 487
  73. See Eric Goldman, 2024, Generative AI] incumbents can use industry regulation to raise new entrants' costs and deepen their competitive moat, Generative AI Is Doomed, 13
  74. Anthropic Commercial Terms of Service I.1, 34
  75. 2024, AI's Big Shift from 'Model-Forward' Innovation to 'Product-Forward, FORTUNE
  76. R Patrick, Moral Reflections on Strict Liability in Copyright, 44 COLUM, J.L
  77. 2021, ARTS, 123
  78. Oren Bracha, 2016, Copyright Accidents, 96 B.U. L. REV. 1025
  79. Diamantis, 2013, 334 RESTATEMENT (THIRD) OF AGENCY � 1.01. 335 Hollingsworth v, 693
  80. Aslihan Asil, 2010, The Criminal Liability of Artificial Intelligence Entities -from Science Fiction to Legal Social Control, 4 AKRON INTELL, Can Machines Commit Crimes Under U.S. Antitrust Laws?, 3 U. CHI. BUS. L. REV, 1
  81. Id, 348 In addition, joint and several liability is not a viable alternative. It would present the same issue of de facto liability, albeit at a lower damages threshold for the developer or user than if they were solely liable. Regardless, joint and several liability does not fit the facts of AI-generated infringements. The developer and user are typically not acting in concert with the AI system to achieve the infringement or independently causing an indivisible injury. See RESTATEMENT (THIRD) OF TORT: APPORTIONMENT OF LIABILITY � � 12, 15, 18A. Negligence-based joint and several liability may seem more attractive, but assigning liability between defendants still considers "the strength of the causal connection between the person's risk-creating conduct and the harm
  82. E G See, RIAA Settles First Lawsuit Against 12-Year-Old Brianna
  83. Technewsworld Lahara, 2003, Eric Bangeman, "I Use Dead People�, ARSTECHNICA, 3
  84. 1004, 363 Perfect 10, 673
  85. Goodyear, 2007, Inc. v. Visa Int'l Serv. Ass, 10, 13
  86. E G See, 2016, YouTube did not decline to exercise 382 See supra Part IV.A. 383 EMI Christian Music Grp, 3
  87. at 1153 (examining how the unpredictability of AI raises material normative problems for applying direct or vicarious liability)
  88. Marchisio, at 11 (noting the need to shift the discussion about obligations away from producers and developers when robots can act autonomously from their original design)
  89. Teubner Beckers, 2007, rejecting one-size-fits all approaches to AI liability and proposing looking at the specific use of the AI system and the involvement of the human, Inc. v. Visa Int'l Serv, 670, 94
  90. See Bracha, supra note 254, at 49 (proposing negligence for AI-generated outputs)
  91. F, 3d at 671. For a discussion about these different standards
  92. Cooper, One could also engage in structural removal, which can achieve a rough approximation of retraining from scratch, but this technique would still be very expensive and has largely been explored with simpler traditional AI rather than large machine learning-based generative AI systems, 8
  93. See Dan, 2019, Algorithmic Fair Use, 86 U. CHI. L. REV, 283
  94. Eugene Volokh, 2023, Freedom of Speech and AI Output, 3 J. FREE SPEECH L. 651, 654, 658
  95. Cass R Sunstein, 2023, But see Dan L. Burk, Asemic Defamation, or, the Death of the AI Speaker, 22 FIRST AMENDMENT L. REV, 189, 228
  96. N Peter, 2024, AI Outputs Are Not Protected Speech, L. REV, 83
  97. See supra notes 312-316 and accompanying text
  98. 2007, 1995)) (contributory liability would lie where a service provider fails to take "simple measures to prevent further damage to copyrighted works, quoting Religious Tech. Ctr. v. Netcom On-Line Commn's Servs., Inc., 907 F. Supp. 1361, 1375, 729
  99. Dr, ComicMix LLC, 983 F.3d 443, 458
  100. 1984, The two respondents in this case do not seek relief against the Betamax users who have allegedly infringed their copyrights, 133
  101. James Grimmelmann, 2023, An Economic Model of Intermediary Liability, 37 BERKELEY TECH, L.J, 1011
  102. Introducing CopyrightCatcher, supra note 115
  103. E G See, 2005, 913, 934

Metadata

Authors
Goodyear, M.
Affiliations
New York University (US)
Corresponding author
Goodyear, Michael
Venue
SSRN Electronic Journal
Publisher
Elsevier BV
Published
2025
Pages
26-38
OpenAlex
W4408395335
Cited by
2
References
55
Citations by year
2026: 1 · 2025: 1
FWCI
1.90
Type
conference paper
Language
en
Open access
Locations
SSRN Electronic Journal (acceptedVersion)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Law, AI, and Intellectual Property
  • Social Sciences › Law › Intellectual Property Law
  • Social Sciences › Law › Intellectual Property Rights and Media
Keywords
generative AI, copyright liability, direct liability, causation requirement, computer-generated works, copyright doctrine, fair use, compulsory licensing, secondary liability, technological change and copyright

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

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.