Germany's #Bundesgerichtshof (Federal Court of Justice) is set to refer questions about training of systems using copyrighted materials to the European Court of Justice
linkedin.com
#copyright #ai #cjeu | Richard Dissmann
Today at the Bundesgerichtshof: LAION goes to Luxembourg
I had another excursion to the German Federal Court of Justice (“Bundesgerichtshof” – “BGH”) today with my colleague Laura Jones (pictured right), in a matter on which I will report under separate cover. Our hearing was scheduled for 10 am, but then it was pushed back to 12 noon for another case which the BGH apparently thought was more important than ours: the famous “LAION” case on data mining and copyright.
LAION e.V., a private law association (“Verein”) under German civil law, claims to be a non-for profit research organization. They created a database of roughly 5,8 billion data sets with links to picture files on the internet. LAION checked if the information on the contents of these pics contained in their metadata was accurate, i.e. whether they really describe what is on the pics. For that purpose, the pictures were copied and thus “reproduced”.
The core question of the case is whether that reproduction of the copyrighted files is covered by the “data mining” exceptions under Artt. 3 and 4 of the Directive on Copyright in the Digital Single Market (“CDSM Directive”).
The BGH said that it will refer the case to the CJEU for preliminary rulings on a number of questions. Based on today’s hearing, these question will probably include the following:
1. Is what LAION does “data mining” at all? The argument against it is that they do not “mine” anything, but just verify whether the metadata in 5,8 billion pics are correct or not (and put the correct ones in a new table).
2. Re Art. 4 of the Directive, if the right owner wants to make an “expressed reservation” against data mining, how does that have to be designed? Must it be explicit, or can it be derived from the context?
3. What does it mean that the reservation must be “machine readable”? Does “machine readable” mean “machine understandable”?
4. Re Art. 3 of the Directive, what are “research organizations”, and what is “scientific research”? Can this also be a private association like LAION?
5. Last but not least, what impact does it have on the “research organizations” and their “scientific research”, if other commercial organizations (like e.g. big tech companies) have a relevant influence on them (e.g through their people sitting on the board of the “research organization”).
At the heart of the case, the real questions are of course whether and in which way the big tech and AI companies can benefit from the data mining exceptions under EU law.
The BGH will announce its decision on 17 December, but based on today’s hearing (and not surprisingly), it is clear that a referral will be made to the CJEU, so LAION will go to Luxembourg now – from Karlsruhe with love.
Our own case was sidelined a bit by the attention given to the LAION case, but we will report on it when there is a result. In the meantime, you can take the clues from the picture to guess what our case was about.
#Copyright #AI #CJEU
