EU’s ‘AI Act’ Moves Another Step Closer to Passage With New Vote — ‘We Want Content That Is Produced by AI To Be Recognizable As Such’

  • Save

  • Save
Photo Credit: Waldemar

A little over one month after the formal approval of negotiations on the EU’s AI Act, the regulatory measure has moved another step closer to becoming law. And while the legislation’s precise music industry impact remains to be seen, the bill as written would ban certain AI applications altogether.

The European Parliament just recently announced that it had “adopted its negotiating position on the” AI Act – specifically with 499 votes in favor, 28 votes against, and 93 abstentions – “ahead of talks with EU member states on the final shape of the law.” Plus, the EP in an overview highlighted some of the newest additions and changes to the voluminous two-year-old bill.

On the latter front, the legislation’s “high-risk AI” category has once again expanded, per the summary, referring particularly to “recommender systems used by social media platforms (with over 45 million users).” Needless to say, Twitter, Facebook, TikTok, and Instagram all boast far more than 45 million users apiece and have each grappled with EU fines and/or regulatory scrutiny as of late.

Regarding the outright “bans on intrusive and discriminatory” practices, these AI uses at present encompass “remote biometric identification systems” in public, “emotion recognition systems in law enforcement,” and the “untargeted scraping of facial images from the internet or CCTV footage to create facial recognition databases,” among others.

Closer to the music space, generative AI systems “like ChatGPT” would under the AI Act be compelled “to comply with transparency requirements” – including “disclosing that the content was AI-generated” and publicly releasing “copyrighted data used for their training,” according to the text.

The stipulations have entered the AI Act’s draft and the media spotlight about 10 days after European Commission VP for values and transparency Věra Jourová called for labels on AI content. And in a press conference announcing today’s vote, MEP Brando Benifei drove home the point and his commitment to making the sweeping legislation law.

“We need to be clear that no voluntary initiative,” Benifei spelled out, “no global effort to coordinate – which is crucial, I said it at the beginning – will hinder or influence in a limiting way the work that we are doing to have strong legislation (especially on transparency in this case) when we deal with generative AI. We want content that is produced by AI to be recognizable as such.”

With AI playing an increasingly significant role in the music sphere – as unauthorized soundalike releases are debuting alongside non-infringing tracks – it’ll be particularly interesting to see what the requirement means for streaming services like Spotify. (Lobbying data shows that the Stockholm-headquartered platform has held multiple high-level European Commission meetings in 2023.)

As it stands, though, evidence and history suggest that the AI Act is a ways off from becoming law and going into effect. Of course, artificial intelligence’s rapid evolution isn’t showing any signs of slowing, with a growing list of all-time-great artists now incorporating the unprecedented technology into the creative process. Stateside, some members of Congress have expressed interest in passing legislation that would establish a regulatory framework for AI.