Will OpenAI be lawful in the EU?
One of the provisions of the AIA is that providers of general purpose AI systems – like OpenAI’s LLM – must “put in place a policy to respect Union copyright law in particular to identify and respect, including through state of the art technologies, the reservations of rights expressed pursuant to Article 4(3) of Directive (EU) 2019/790”.
?Directive (EU) 2019/790 is the Copyright And Related Rights Directive and one of the things it does is to allow web-scraping (aka text and data mining) for the purposes of scientific research, but – and this is the important point – it requires the permission of rightholders for all other uses.
?You might think that this only applies to the EU rights of rightholders so, if OpenAI had trained its LLM only on US data, there would no issue. But one of the recitals of the AIA provides:
?“Any provider placing a general purpose AI model on the EU market should comply with this obligation [i.e. right-holder consent], regardless of the jurisdiction in which the copyright-relevant acts underpinning the training of these general purpose AI models take place. This is necessary to ensure a level playing field among providers of general purpose AI models where no provider should be able to gain a competitive advantage in the EU market by applying lower copyright standards than those provided in the Union.”
?It's clear from the New York Time v OpenAI litigation that OpenAI did not get the NYT’s consent to use its articles as training data. And, it looks like the wording of the AIA means that, even if OpenAI won its case on fair use, it still would not meet the threshold set by the AIA.
Will OpenAI be lawfully usable in the EU once the AIA comes into force? Mmmmm…..looks doubtful.
See also the next edition of AI Legal on the extra-territorial effects of the AIA.
Cutting Edge Lawyering for B2B SaaS
10 个月Fabio Ronca Hi Fabio, thanks for your comment. You are making a good point. In the old days, it was hard work to copy someone’s copyright work at scale, and the only people that did copy at scale were the obvious bad guys (eg.DVD pirates).?Nowdays, it’s easy to copy other people’s work on an industrial scale and so the question arises whether copyright protection should be adjusted to reflect this.? Having said this, the EU is the most advanced legislator in the world for digital related activities, and the French competition authority has just fined Google 250m for infringements by Bard in using the articles of French publishers as part of its training data. In the US, as I’m sure you know, the New York Times is suing OpenAI and Microsoft for having copied 200 years’ worth of articles as part of training OpenAI. So, things are moving.?But it’s going to be slow and painful. I will be covering the Bard fines, NYT v Open AI and other related issues in forthcoming issues of AI Legal. But please let me know if there are specific issues you would me to cover. Also in forthcoming issues: the AI Act in bite-sized chunks.
Manager Organizational Development | Information and Communications Technology, SAP S/4HANA | Inoovation and transformation manager
10 个月Mark, you had a brilliant idea with this group and its newsletter, it seems necessary to better understand how AI can be used while maintaining the rights of those who have put their ingenuity into play. I remain very doubtful about the control of these rights and, despite having started a jurisprudential path, I am still not clear on effective and efficient control structures and methods. Isn't it that a beautiful law was made but then no one checks?
Great idea Mark!