AI training and copyright: can it be done lawfully by relying on provisions other than Article 4 DSMD, for example research- and education-specific exceptions or fair use defences? I answer this in the negative in a study just published in @EurJRR

Here's the abstract of the article (published online in autumn 2024) All feedback welcome! Thanks to @IFPI_org, the anonymous reviewer, and the whole @EurJRR Team and @alemannoEU for hosting my work 2/2

The latest issue of @EurJRR has just been released and also features my discussion of liability aspects facing AI-generated outputs under copyright and related rights. This is an area whose relevance appears to have grown considerably over the past few months 1/2.

A pre-print of the article is available on SSRN 4/4.

Both questions are answered in the negative: no exception/open-ended defence fully covers unlicensed AI training activities. As a result, a licensing approach appears to be the way for AI training to be undertaken lawfully, including when done for ‘research’ and ‘learning’. 3/4.

This new research article for @EurJRR, addresses 2 key questions: (a) if unlicensed AI training can be classified as ‘research’ or even ‘learning’ in the context of ‘teaching’, and (b) if commercial AI developers can take advantage of related exceptions and fair use defences. 2/4.

Can unlicensed AI training be performed by relying, not on Art 4 of Directive 2019/790 (which has been also figuring prominently in the ongoing UK reform debate), but rather on other exceptions, notably those for research and education? And what about fair use doctrines? 1/4.

AG Spielmann has (unsurprisingly) advised the CJEU to rule that a derivative work may be ©️-protected if original, while also noting (more surprisingly) that free and creative choices may be made when one seeks to reconstruct/restore someone else's work

AG Emiliou has delivered his lPelham II Opinion. While the proposed definition of 'pastiche' is broadly acceptable - including the refusal to consider it a 'catch-all' concept - parts of the analysis are not relevant to the referred questions

The CJEU has received first referral specifically on AI and ©️. A 🇭🇺 court has asked if: (1) A chatbot engages rights of reproduction/making available to the public; (2) AI training is relevant under reproduction; and, if so, (3) The TDM exception(s) apply

The centrality of originality is unsurprising, and in my view should continue informing the infringement test too (without any recognizability requirement) 2/2

This morning, AG Szpunar issued his long-awaited Opinion in Mio/konektra: works of applied art are protected by copyright subject to an originality requirement, which is all that matters. BUT: the test of infringement should be based on recognizability 1/2.

AG Campos has advised CJEU to rule that joint ©️ ownership is a matter governed by national, not EU law. While formally correct, the Opinion might move from an incorrect premise, that is that 'author' and 'rightholder' are not in fact also harmonized

RT @FIMI_IT: Al via il convegno di @MiC_Italia sulle opportunità, i rischi e la sfida nella regolamentazione dell'AI nell'industria musical….

RT @FIMI_IT: #SaveTheDate • Le opportunità, i rischi e la sfida nella regolamentazione dell'AI nell'industria musicale nell'analisi delle a….

The US Court of Appeal for the DC Circuit has confirmed that human authorship is a requirement for protection under US copyright. An unsurprising outcome, which indicates how the real issue is not *who* an author is but rather *what* makes one an author

For those interested, if you get the book on the OUP website and use code AUFLY30, a 30% discount shall be applied

Huge thanks to @OUPLaw for giving a home to this project and to ALL the several contributors who have shared their expertise in this emerging and fascinating area of the law. 2/3.

👠 👗 🎩 FASHION (LAW) ALERT! The Handbook of Fashion Law, which I have had the pleasure of editing together with @Ire_Calboli, is finally here! 1/3

'MAGIC PUSSY' for coffee (whether virtual or actual one) is not a trade mark contrary to "accepted principles of morality". A sensible decision of the EUIPO 5th BoA, though a not-so-magic question remains: how helpful is actually Article 7(1)(f) EUTMR?