IPSC 2020 at Stanford: Mauritz Kop presented Machine Learning & EU Data Sharing Practices at the Intellectual Property Scholars Conference — the works-in-progress forum where IP scholarship is stress-tested before journals see it. The 2020 edition, hosted by Stanford Law School, ran as virtual panels from July 15 through August 5 in the first pandemic summer.
Training data under four regimes at once
Machine learning is hungry, and in Europe its raw material sits under copyright, database rights, trade secrets and the GDPR simultaneously. The paper mapped that intersection — including the text-and-data-mining exceptions of the DSM directive — and asked which data-sharing arrangements actually let lawful European AI development proceed at scale.
An argument for coordination
Where exclusive rights and data-protection rules overlap without coordination, they tax exactly the data flows the EU's own artificial intelligence strategy depends on. That modernization argument, workshopped before a predominantly American IP audience with a different copyright baseline and fair-use culture, had to hold up under comparative fire — which is precisely what the IPSC format is for.
Part of the Stanford research agenda
The presentation belonged to Kop's research line at the Stanford-Vienna Transatlantic Technology Law Forum, which he had joined earlier that year — see Mauritz Kop becomes TTLF Fellow at Stanford University. The paper is preserved in the permanent Stanford RQT collection at the Stanford Law Library, and its data-protection companion piece appeared in the Harvard Journal of Law & Technology's digest — two halves of one question about Europe's machine-learning data rules.
A format built for critique
Short presentations, dense Q&A, no published proceedings: IPSC exists purely to make drafts better before journals see them. For interdisciplinary work spanning artificial intelligence, data governance and IP doctrine, an audience of doctrinalists, economists and technologists probes each weak point in turn — and a European paper before an American room must hold up under a different copyright baseline and fair-use culture besides.
Why it still matters
The training-data questions posed in that 2020 draft — who may train on what, and on which terms — have since moved to the center of AI regulation on both sides of the Atlantic. Opt-out patchworks under the text-and-data-mining exceptions, the GDPR's reach into model pipelines, the competitive pull of jurisdictions with cleaner data rules: each was on the table at that virtual Stanford panel before it reached the regulators' agenda. The workshop critique of that summer became part of the foundation the later debates built on — which is exactly what a works-in-progress conference is supposed to produce.
Meer lezen