IFPI, GESAC and others call for mandatory disclosure of artificial intelligence training in the EU – “AI innovation and strong copyright protection are not mutually exclusive”

  • Save on computer

IFPI GESAC and others call for mandatory disclosure of artificial | Richmcculley

  • Save on computer

Brussels Belgium. Credit: Francois Genon

Amid running trilogue negotiations With regard to the EU’s AI law, music industry organizations including IFPI, IMPALA and GESAC are calling on lawmakers to ensure that legislation compels artificial intelligence systems “to comply with the existing EU copyright framework”.

These entities, along with the ICMP, IMPF and a number of non-musical media representatives, collectively called for “meaningful transparency obligations for AI” in a short notice emailed to DMN.

This latest push to introduce clear copyright protections into AI law comes at a time when the law is fast approaching passage – and a growing number of creators taking legal action against AI developers for allegedly “training” their platforms on protected media without permission.

“Advances in AI innovation and effective copyright protection are not mutually exclusive,” write the signatories in their message to the AI ​​Act. “On the contrary, AI processes that rely on the ‘input’ of protected works or other subject matter derive their purpose and value from those works or subject matter.”

Building on this, IFPI and the other bodies mentioned above expressed their position that the companies behind AI offerings should be required under the AI ​​Act to “keep detailed records of the works of third parties or other subject matter used”.

This information, the organizations say, should also be accessible to rightsholders, especially so that they can “enforce their rights effectively” when protected works are used without permission.

“The commitment to keeping accurate records should go back to the beginning of development to ensure a full chain of use,” the media and entertainment officials wrote. “Furthermore, in order to avoid ‘AI laundering’, it must extend to all systems made available or the results of which are used in the EU, regardless of the jurisdiction in which the training or examination took place.”

Taking these comments into account, IFPI, ICMP, IMPF, GESAC and IMPALA concluded that the European Parliament’s recent proposal to keep records of AI training was “a first step in the right direction”.

“We call on the European institutions to support these provisions and we look forward to working with them to make further improvements to ensure that AI law is fit for protecting the work of European authors and rightholders,” they concluded.

Of course, artificial intelligence is playing an increasingly important role by the day — especially in the music space, where unauthorized sound-alike tracks (and entire streaming platforms focused on AI efforts) are getting a lot of attention.

Authorized AI music services are also on the rise (Mubert announced this month that its catalog has surpassed the 100 million song mark), and companies including but not limited to TikTok parent company ByteDance have recently been adopting AI-powered music generators.

With this in mind, major labels are seriously considering streaming fee reform and are urging US lawmakers to introduce tougher AI protections.

Additionally, in July, Epidemic Sound introduced “an AI-powered music discovery tool” called Soundmatch, and K-pop artist Mark Tuan launched an AI “digital twin” that can interact with fans. Meanwhile, Aimi uses artificial intelligence to generate trap music, while BMG partner MatchTune has developed a self-described “natural language search engine for music” called Audioatlas.