In November 2023, Eurojust published a new report analysing the impact of Generative Artificial Intelligence (AI) on intellectual property crimes. Generative AI refers to a category of artificial intelligence techniques and models that are designed to generate new and original content, including text, imagery, audio, and other data.

Recent examples of such generative AI are technologies such as ChatGPT, DALL-E, and Bard. The development of generative AI is further supported by (1) the development of so-called large-language models (LLM) that can perform language processing tasks and (2) multimodal AI that can recognise various types of data, including text, speech, videos, and images at the same time. As a result, generative AI can create different types of content, music, and images, which gives rise to numerous questions for the copyright system and the law; as it stands today, they protect works that are original, with most definitions of originality requiring a human author.

Raising the question of whether generative AI can infringe intellectual property (IP) rights and whether it can produce IP-infringing results, the report first explains the generative AI "training" process, which basically involves feeding it with massive amounts of publicly available data. If this data includes copyrighted works, the question arises as to whether their use is permissible. The report gives a short overview of the answers given by US courts, which are currently dominated by the fair use doctrine, and courts in the EU Member States, which seem to take a stricter approach under the EU’s Directive (EU) 2019/790 on copyright and related rights in the Digital Single Market.

Proceeding from there, the report addresses the question of whether AI produced material can be transformative enough not to pose a threat to the creators or compete against their work. Such situations are expected to raise many legal issues that will need to be addressed by the courts in future. One such possible scenario: an AI machine-generated song based on the lyrics and music of existing songs composed by many different artists.

AI-generated answers, for instance by ChatGPT, are used to illustrate how these applications work. Lastly, the report analyses how generative AI can be used by criminals to advance their modi operandi. Here, the report finds and explains several new modi operandi:

• The creation of cybercriminals' own malicious generative AI;
• The employment of generative AI to advance the unlawful streaming of copyright material;
• The use of generative AI in counterfeiting, the creation of counterfeit products, and the identification of grey markets;
• The infringement of trade secrets by using generative AI to construct malware;
• Trademark registration invoice fraud, where generative AI can be used to develop false invoices, emails, and communication papers and even generate logos like actual IP registration agencies.

The report was prepared by the Intellectual Property Crime Project funded by the European Union Intellectual Property Office (EUIPO) and executed by Eurojust.