Editing Openai/6924739c-2300-800e-9743-65c152c213f1 (section)

=== Assistant: Here are the foreign authorities that, in my view, are the most significant and directly relevant to the issues in the … ===
Here are the foreign authorities that, in my view, are the most significant and directly relevant to the issues in the ROSS / Thomson Reuters dispute (copyright in headnotes / digests, intermediate copying for AI training, and “research”/TDM-style exceptions).

I’ll group them by theme and flag the ones that are especially high-value for this case.

==== ### ====

Why it matters here
* The Court held that headnotes, case summaries, topical indexes and the compilation of reported decisions are original works in which copyright subsists.
* It also set the “skill and judgment” test for originality (rejecting pure “sweat of the brow”), and developed a structured fair dealing framework (purpose, character, amount, alternatives, nature of work, effect).

Relevance to this case
* Direct analogue to Westlaw: CCH explicitly recognizes copyright in law-reporter headnotes, summaries, and indexes – the very sort of editorial content at issue here.
* Its fair-dealing analysis (especially the emphasis on “user’s rights” and technological neutrality) is a natural comparative reference point for U.S. fair use arguments about copying headnotes to support legal research tools.

===== What it held =====
* EBC publishes Supreme Court Cases (SCC) law reports. Defendants produced CD-ROM law databases alleged to have copied SCC’s materials.
* The Indian Supreme Court: - Agreed that headnotes, editorial notes, and some formatting/structural edits are copyright-protected; - Rejected copyright in the raw judgment text itself (public domain), but held that some copy-editing and structuring can reach the originality threshold; - Adopted the CCH “skill and judgment” standard, rejecting both pure sweat-of-the-brow and a high creativity threshold.

Relevance
* Factually very close: a law-report publisher vs. a digital legal research/database provider accused of using the publisher’s edited judgments and headnotes in electronic products.
* The court drew a line between public-domain judicial text and copyrightable editorial enhancements (headnotes, footnotes, paragraphing, introductions), then limited what could be freely reused.
* That structure maps closely onto arguments about what parts of Westlaw content ROSS could ingest, and whether copying for a competitor research platform is infringing.

===== What happened =====
* Reporters provided case reports for The Jurist; each report had a headnote summarizing the decision.
* The defendants published a monthly digest which reproduced headnotes/marginal notes from various reports (including The Jurist).
* The court held the plaintiffs had copyright in the reports, and the copying of the headnotes into the digest amounted to piracy.

Relevance
* Classic authority on copyright in law-report headnotes and on copying those headnotes into a separate digest product.
* Strong comparative support for Thomson Reuters’ side that headnotes and classification-style editorial content are protectable and can’t simply be taken into a competing system.

===== - Eastern Book Co. v LexisNexis injunction: later Indian proceedings confirmed injunctive protection for EBC’s law-report content against a competing legal database. =====
* David Vaver, “Copyright Inside the Law Library”: summarises Canadian law that publishers can object to copying of their headnotes, digests, and cumulative indices such as those in Dominion Law Reports.

These are secondary or follow-on, but they reinforce a transnational pattern: courts routinely protect law-report editorial matter separate from public-domain judgments.

==== ### ====

Key holdings
* An 11-word extract of a protected work can be a “reproduction in part” if it embodies the author’s intellectual creation.
* Printing such snippets during a data-capture process is not a transient copy under InfoSoc Article 5(1); consent is required.

Relevance
* Shows how EU law treats intermediate copying in text-processing and data-capture workflows – very close to what happens in text and data mining or AI training.
* Supports the proposition that even small fragments of headnotes or key-number text can be protected, making wholesale ingestion for training a legally significant reproduction under EU standards.

===== Key points =====
* Concerned copyright in football fixture lists as databases.
* The CJEU held that a database is protected by copyright only if the selection or arrangement of data is the author’s own intellectual creation; investment in obtaining/checking data (“skill and labour”) alone isn’t enough.

Relevance
* Analogue for West’s key-number system and editorial classification: - To the extent U.S. courts are wrestling with how much originality is in a systematized legal digest, Football Dataco offers a fully worked-out “intellectual creation” vs mere labour framework.
* Also underscores that some structural, taxonomic choices in a legal database can be protectable, even when the underlying case law isn’t.

==== ### ====

C-360/13 (CJEU, 2014) and [2013] UKSC 18

What it decided
* Considered whether online media-monitoring services and their clients infringed copyright by receiving and viewing snippets/links.
* The CJEU and UKSC held that the on-screen and cached copies made in ordinary web browsing fall within the InfoSoc “transient and incidental copy” exception and don’t require a licence.

Relevance
* Offers a nuanced foreign treatment of automated technical copies vs. more enduring reproductions in digital services.
* For ROSS-type systems, this line of authority helps frame arguments that some system-level copies are purely technical (and thus potentially excusable) while full-scale ingestion for training may go well beyond that transient category.

==== ### ====

What the Directive does
* Introduces two TDM exceptions: - Art. 3 – mandatory exception for research organisations/cultural heritage institutions doing TDM for scientific research. - Art. 4 – broader exception allowing anyone to perform TDM unless rightsholders opt out using machine-readable reservations.
* Scholarship and policy analyses confirm these provisions were intended to cover automated analysis and AI training on large corpora.

Relevance
* This is the leading legislative model for handling AI-style training by exception rather than by fair use.
* It would have been a natural comparative touchstone for: - Arguing that when legislatures want to bless AI training on protected content, they do so expressly via TDM exceptions; - Or conversely, that absent such an exception, broad unlicensed training remains presumptively infringing.

===== What happened =====
* Photographer sued LAION over use of images scraped from the web and incorporated into an AI training dataset.
* The Hamburg court held that reproductions for training could fall under Germany’s TDM exception for scientific research (Section 60d UrhG, implementing DSM Art. 3), and that AI training counts as TDM.

Relevance
* One of the first court decisions squarely addressing AI training and TDM in Europe.
* Shows a judicial approach that: - Accepts AI training as TDM; - Uses statute-based exceptions rather than judge-made fair use; - Still leaves room for rightholder opt-outs and limits (especially when outputs reproduce works).

A U.S. court could look to this as a contrast model: in the EU, statutory exceptions + opt-outs are doing work that in the U.S. is being asked of fair use.

===== Key features =====
* Japan’s 2018 amendment created Article 30-4, which permits the use of copyrighted works for “information analysis”, including AI development and training, without needing authorisation, subject to certain conditions (for example, not using the works to enjoy them as such, and guarding against certain market harms).

Relevance
* Another major economy that has explicitly legalised AI training on copyright works via a broad TDM-style exception.
* Very useful as comparative material to show that where a jurisdiction wants to ensure AI training is lawful, it does so affirmatively—again reinforcing the idea that in the U.S., courts are being asked to fill a policy gap via fair use.

===== What it says =====
* Section 29A allows making copies of works for text and data analysis for non-commercial research, provided the user has lawful access to the work.
* There has been active UK discussion about broadening this into an EU-style TDM exception that would more clearly cover AI training.

Relevance
* For Westlaw-style subscription access, this is particularly on point: - It suggests a model where lawful subscribers can data-mine content in limited research contexts without permission, but commercial exploitation (like building a competing product) remains contested or excluded.
* It’s a neat comparative frame for where ROSS-type uses might sit if this were litigated under UK law.

==== These are less about headnotes specifically but very relevant to fair use / fair dealing analysis of AI training. ====

===== - The Supreme Court held that 30–90 second music previews offered by online stores were fair dealing for the purpose of “research” under s. 29 of the Copyright Act, even in a commercial setting. =====
* The Court emphasised: - A “large and liberal” interpretation of “research”; - That fair dealing is a user’s right; - The importance of technological neutrality.

Relevance
* Supports arguments that automated, intermediate copying that enables users’ research (like training a legal research AI) can be consistent with a robust user-rights conception, especially where: - Outputs don’t substitute for the original work; - The dealing may promote access to law and legal information.

===== - Clarified that downloading a digital copy engages the reproduction right but not a separate “communication to the public” right, highlighting the principle of technological neutrality in Canadian copyright. =====

Relevance
* Not about TDM per se, but supports a comparative approach that looks at the substance of what’s happening (creating copies for a new use) rather than the technical form — helpful analogical material for fair-use arguments about intermediate copies during AI training.

==== If you only had room for a handful of foreign authorities that are both substantively on-point and likely to be treated as serious comparative law by a U.S. federal court, I’d prioritise: ====
# CCH Canadian Ltd v Law Society of Upper Canada (Canada) – for both - copyright in headnotes / law-report enhancements, and - a structured, user-rights-oriented fair dealing framework.
# Eastern Book Company v D.B. Modak (India) – nearly a factual cousin to Thomson Reuters v ROSS: law report publisher vs digital database; headnotes and edited judgments; adoption of the CCH originality test.
# Sweet v Benning (UK) – historic but very on-point authority on copying headnotes into a digest, supporting strong protection for law-report editorial content.
# Infopaq (C-5/08) and Football Dataco (C-604/10) (EU/CJEU) – together they give a sophisticated foreign view on: - How small fragments (11 words) can be protected at the reproduction stage; - How selection/arrangement originality in a database is evaluated.
# DSM Directive Arts. 3–4 + Hamburg/LAION decision – to show that when AI training is meant to be privileged, many jurisdictions rely on explicit TDM exceptions (sometimes with opt-outs) rather than stretching general fair-use doctrines.
# Japan’s Article 30-4 – as a contrasting legislative choice to affirmatively permit broad AI training uses of copyrighted works.
# Optionally, SOCAN v Bell Canada – if you want a common-law Supreme Court decision aggressively expanding “research” and user rights in a technologically complex, commercial context.

If you tell me which side of the case you’re supporting (ROSS or Thomson Reuters), I can help narrow this down further into “good for us, bad for them” buckets and suggest how each authority could be framed rhetorically.