Unjust Enrichment as a Remedy for AI's Unauthorised Use of Protected Data

Jyh-An Lee; Yangzi Li

arxiv: 2605.23503 · v1 · pith:WGFPGMICnew · submitted 2026-05-22 · 💻 cs.CY

Unjust Enrichment as a Remedy for AI's Unauthorised Use of Protected Data

Yangzi Li , Jyh-An Lee This is my paper

Pith reviewed 2026-05-25 02:52 UTC · model grok-4.3

classification 💻 cs.CY

keywords unjust enrichmentAI training datagenerative AIdata ownershiplegal remediesrestitutionIP infringementprivacy law

0 comments

The pith

Unjust enrichment doctrine provides a remedy for unauthorized AI data use by recovering benefits rather than proving wrongdoing.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper argues that unjust enrichment can serve as an alternative to IP and privacy laws when addressing unauthorized data use in training generative AI models. It claims this doctrine captures the wrongfulness of such use in a distinct way, allowing data owners to recover benefits gained by developers without needing to establish fault or infringement. The approach is presented as more pragmatic because it shifts focus from wrongful conduct to unjust retention of value. If correct, this would reconcile data owners' interests with AI innovation more effectively than existing legal, equitable, or statutory remedies.

Core claim

Unjust enrichment captures the wrongfulness of unauthorised data use in a manner distinct from IP infringement and privacy violations. Gain-based restitution for unjust enrichment may prove more advantageous than existing remedies. By shifting the emphasis from establishing wrongful conduct to recovering benefits obtained unjustly, unjust enrichment offers a pragmatic and equitable framework that reconciles the rights of data owners with the interests of AI developers.

What carries the argument

The doctrine of unjust enrichment, which enables recovery of benefits obtained unjustly from unauthorized data use in AI training.

If this is right

Data owners could recover the value AI developers obtained from their data without proving infringement or violation.
AI developers could face restitution obligations even in the absence of proven wrongful conduct.
Gain-based remedies could supplement or partially replace damage calculations under IP and privacy statutes.
Disputes over data ownership in AI training could resolve through equitable balancing rather than fault-based litigation.
The framework could encourage voluntary compensation arrangements between data holders and AI trainers.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

This remedy might extend to other machine learning applications that scrape public or protected data at scale.
Courts may need new methods to quantify the 'benefit' received when data contributes to model weights or outputs.
Widespread adoption could pressure AI companies to document data sources more rigorously to defend against restitution claims.
The approach could interact with emerging data licensing markets by providing a fallback when contracts are absent.

Load-bearing premise

Unjust enrichment can be applied to unauthorized data use in AI training in a way that is meaningfully distinct from IP infringement and privacy violations, with gain-based restitution proving more advantageous than current remedies.

What would settle it

A court ruling or detailed case analysis demonstrating that unjust enrichment claims for AI data use either collapse into IP or privacy claims or produce remedies no better than existing options.

read the original abstract

The unauthorised use of data in the training of generative AI models presents significant legal challenges, particularly under intellectual property (IP) and privacy laws. These frameworks frequently grapple with the intricate relationship between data ownership and AI innovation, resulting in ongoing debates regarding optimal protection and enforceability. This article delves into considerable potential of unjust enrichment as an alternative legal doctrine for resolving disputes arising from such unauthorised data use. We explore how the concept of unjust enrichment captures the wrongfulness of unauthorised data use in a manner distinct from IP infringement and privacy violations. Furthermore, we analyse the extent to which gain-based restitution for unjust enrichment may prove more advantageous than existing remedies, including legal, equitable, and statutory options. We content that by shifting the emphasis from establishing wrongful conduct to recovering benefits obtained unjustly, unjust enrichment offers a pragmatic and equitable framework that reconciles the rights of data owners with the interests of AI developers.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper floats unjust enrichment as a gain-focused alternative to IP and privacy claims for unauthorized AI data use, but stays at the level of suggestion without much precedent or comparison.

read the letter

The core pitch is that unjust enrichment lets data owners recover benefits AI developers obtained without needing to prove traditional wrongdoing, which the authors say sidesteps some of the ownership and enforcement headaches in current law. That framing is the main takeaway for anyone thinking about remedies in this space. The paper does a reasonable job identifying why IP infringement and privacy rules often leave gaps when data is scraped for training: they tie up in questions of ownership and intent that don't always fit bulk data use. Shifting the lens to unjust enrichment keeps the focus on the benefit received, which could make recovery simpler in some cases. It also tries to balance interests by noting that this approach might be less disruptive to AI development than full injunctions. The soft spot is the lack of depth. The abstract and summary give no specific cases, no jurisdiction-by-jurisdiction analysis, and no side-by-side look at how gain-based restitution would actually outperform damages, disgorgement under other doctrines, or statutory options. Without that, the claim that it is meaningfully distinct and more advantageous reads as plausible but untested. The argument rests on standard legal concepts rather than anything circular or invented, which is fine, but it leaves the practical payoff unclear. This is for legal academics and policy people working on AI data governance who want to explore remedy design. A reader already familiar with unjust enrichment in commercial contexts would see the extension quickly. It deserves a serious referee because the topic is live in litigation and the proposal is coherent on its own terms, even if it needs more case law and comparative work to strengthen it.

Referee Report

2 major / 0 minor

Summary. The manuscript proposes unjust enrichment as an alternative legal doctrine for addressing unauthorized use of protected data in training generative AI models. It claims this approach captures the wrongfulness of such use distinctly from IP infringement and privacy violations, and that gain-based restitution may prove more advantageous than existing legal, equitable, and statutory remedies by shifting focus from wrongful conduct to recovery of unjust benefits, thereby reconciling data owners' rights with AI developers' interests.

Significance. If supported by detailed doctrinal analysis, the proposal could contribute to AI governance debates by offering a gain-focused remedial framework. However, the absence of specific precedents, jurisdiction-specific applications, comparative remedy evaluations, or case illustrations in the manuscript substantially limits its significance and practical utility for the field.

major comments (2)

Abstract: The central claim that unjust enrichment 'captures the wrongfulness of unauthorised data use in a manner distinct from IP infringement and privacy violations' and 'may prove more advantageous' is asserted without any supporting legal analysis, cited precedents, or comparative evaluation of remedies, leaving the normative position without visible grounding in the manuscript.
Abstract: No discussion is provided of how unjust enrichment doctrine would apply to data used in AI training (e.g., whether data qualifies as an 'enrichment,' jurisdictional variations in unjust enrichment elements, or interaction with existing data protection statutes), which is load-bearing for the claim of a 'pragmatic and equitable framework.'

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their thoughtful and constructive comments. We address each major comment below and note the revisions we will make to improve the manuscript's clarity and grounding.

read point-by-point responses

Referee: Abstract: The central claim that unjust enrichment 'captures the wrongfulness of unauthorised data use in a manner distinct from IP infringement and privacy violations' and 'may prove more advantageous' is asserted without any supporting legal analysis, cited precedents, or comparative evaluation of remedies, leaving the normative position without visible grounding in the manuscript.

Authors: The abstract summarizes the arguments developed at greater length in the body of the manuscript. We accept that the abstract could more explicitly signpost the doctrinal analysis, precedents, and comparative evaluations contained in the full text. We will revise the abstract accordingly and, if the referee deems it necessary, expand the relevant sections with additional citations and comparisons. revision: partial
Referee: Abstract: No discussion is provided of how unjust enrichment doctrine would apply to data used in AI training (e.g., whether data qualifies as an 'enrichment,' jurisdictional variations in unjust enrichment elements, or interaction with existing data protection statutes), which is load-bearing for the claim of a 'pragmatic and equitable framework.'

Authors: We agree that the abstract does not elaborate on these application details. The manuscript focuses on the conceptual advantages of the framework, but we will add a dedicated subsection addressing the qualification of training data as enrichment, key jurisdictional differences in unjust enrichment doctrine, and the interplay with statutes such as the GDPR and other data-protection regimes. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper advances a normative doctrinal proposal applying established unjust enrichment principles to AI data use, without any self-definitional loops, fitted parameters renamed as predictions, or load-bearing self-citations that reduce the central claim to its own inputs. The argument draws on external legal concepts and precedents for its distinctions from IP and privacy remedies, remaining self-contained as a policy-oriented analysis rather than a closed derivation.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

This is a legal analysis paper. It contains no quantitative models, fitted parameters, mathematical axioms, or invented scientific entities. The ledger fields are therefore empty.

pith-pipeline@v0.9.0 · 5682 in / 1158 out tokens · 28895 ms · 2026-05-25T02:52:33.108197+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

2 extracted references · 2 canonical work pages

[1]

image right

<https:/ /doi.org/10.1038/d41586-024-01578-4> accessed 16 September 2024. 45 ibid. 46 Sara H. Jodka, ‘Manipulating Reality: The Intersection Of Deepfakes And The Law ’ (Reuters, 1 February 2024) <https:/ /www.reuters.com/legal/legalindustry/manipulating-reality-intersection-deepfakes-law-2024-02-01/> accessed 16 September 2024. 47 ibid; see also Quentin J...

work page doi:10.1038/d41586-024-01578-4 2024
[2]

<https:/ /www.nytimes.com/2020/01/18/technology/clearview-privacy-facial-recognition.html>; Kashmir Hill, ‘Clearview AI Used Your Face. Now You May Get a Stake in the Company ’ (The New York Times , 13 June 2024), <https:/ /www.nytimes.com/2024/06/13/business/clearview-ai-facial-recognition-settlement.html> accessed 25 September 2024. 76 Common Law World ...

work page 2020

[1] [1]

image right

<https:/ /doi.org/10.1038/d41586-024-01578-4> accessed 16 September 2024. 45 ibid. 46 Sara H. Jodka, ‘Manipulating Reality: The Intersection Of Deepfakes And The Law ’ (Reuters, 1 February 2024) <https:/ /www.reuters.com/legal/legalindustry/manipulating-reality-intersection-deepfakes-law-2024-02-01/> accessed 16 September 2024. 47 ibid; see also Quentin J...

work page doi:10.1038/d41586-024-01578-4 2024

[2] [2]

<https:/ /www.nytimes.com/2020/01/18/technology/clearview-privacy-facial-recognition.html>; Kashmir Hill, ‘Clearview AI Used Your Face. Now You May Get a Stake in the Company ’ (The New York Times , 13 June 2024), <https:/ /www.nytimes.com/2024/06/13/business/clearview-ai-facial-recognition-settlement.html> accessed 25 September 2024. 76 Common Law World ...

work page 2020