Drawing on intense criticism from online publishers across the European Union (EU) against Generative AI (GAI), the present article aims to highlight the highly debated copyright issue of data collection for Generative AI training. Three questions are therefore addressed: To what extent is scraping data for GAI training considered to be a copyright issue; How Data scraping and data mining are regulated under EU Law and; How the future AI Act intends to deal with the use of training data.

The present contribution provides an overview of two of the most pressing legal questions concerning IP law, authorship of AI-generated works and copyright infringement. It does so through the lens of US copyright law, since US courts have dealt with the highest number of AI art cases, and many AI systems manufacturers and suppliers are based in the US.

On September 7th, 2023, the Court of Justice of the European Union (CJEU) upheld the decision of the General Court according to which the public can partially access documentation on the EU’s emotion recognition project (iBorderCtrl) in which it discusses the general reliability, ethics and legality of such technology.

Eight more American tech companies (Adobe, Cohere, IBM, Nvidia, Palantir, Salesforce, Scale AI, and Stability) signed up to President Joe Biden’s voluntary commitments governing AI (second round of voluntary commitments). In the meantime, a third trilogue will take place on the other site of the Atlantic in relation to the EU AI Act proposal.

The adoption of the negotiating position by the European Parliament sets the stage for the trilogues between the EU institutions, while the European Commission is pushing for the AI Act to be finalised by the end of 2023. The European Parliament’s position on this legislative file reflects its members’ fundamental desire to make the EU a leader in AI regulation and innovation.

On May 17th, 2023, the European Data Protection Board (EDPB) published its final report on the use of facial recognition technologies (FRTs) by Law Enforcement Authorities (LEAs). This report opposes mass surveillance, and, according to the EDPB, ‘the use of facial recognition by law enforcement agencies must be necessary, limited, and proportionate’.

On May 16th, 2023, the French data control agency – Commission Nationale de l’Informatique et des Libertés (CNIL) – published an action plan aimed at ensuring respect for the privacy of people in relation to Artificial Intelligence (AI) systems and more specifically generative AI (e.g. Midjourney and ChatGPT from the company OpenAI). This action plan follows the 2017 CNIL’s first global approach on these new tools.  

Between 21 and 23 April, the European Commission (Commission) held the closing session of the European citizens’ panels, to debate and propose recommendations on virtual worlds in the EU. As a result, a panel of 150 citizens has contributed to the provision of 23 recommendations on ‘fair and human-centric virtual worlds in the EU’.

The European Consumer Organisation (BEUC) is calling ‘for EU and national authorities to launch an investigation into ChatGPT and similar chatbots’, following the filing of a complaint on March 30th, 2023, on the other side of the Atlantic by the Center for Artificial Intelligence and Digital Policy (CAIDP) in relation to ChatGPT-4.

On March 27th, 2023, the European Union Agency for Law and Enforcement Cooperation (EUROPOL) released a report on the Impact of Large Language Models on Law Enforcement. Having considered the outcomes that resulted from a series of workshops organised by the Europol Innovation Lab on how criminals can abuse LLMs such as ChatGPT, as well as how it may assist investigators in their daily work, the report provided useful recommendations on enhancing law enforcement preparedness.

On February 21th, 2023, the United States Copyright Office (USCO) cancelled the registration certificate previously issued to Ms. Kristina Kashtanova for the “comic book” entitled “Zarya of the Dawn”, considering that the images in the comic book are not the product of human authorship – since they were generated by an AI system – Midjourney – and were therefore not copyrightable.

Using innovative AI techniques, Berkeley’s researchers have analysed more than 2.5 million fully anonymized metaverse data recordings and found that individual users could be uniquely identified. The study stresses the need to enhance security and privacy awareness in relation to these platforms.