Aussie songwriters demand AI platforms pay for using their songs for training

Does the world need an AI Jimmy Barnes or Missy Higgins? Aussie songwriters would prefer that didn’t happen and demand AI platforms pay for scraping their music.

Songwriters may dabble with AI for lyrical inspiration but fear tech companies will rob them of more than half a billion dollars in income by scraping their songs for machine learning.

The AI and Music report, commissioned by songwriter’s rights organisation APRA, projects a cumulative damage of $519 million to music creators’ revenues by 2028.

APRA and hitmakers including Josh Pyke, Missy Higgins, Bernard Fanning, Tina Arena and Peter Garrett have called on the Federal Government to act swiftly to force AI companies to reveal the content they are using to train their platforms, seek permission from the copyright owners and pay licence fees for the use of songs.

More than 4200 songwriters and music publishers contributed to the report which revealed 38 per cent had dabbled with AI in their work, for inspiration to create, edit or mix a song or video, or for marketing tools such as press releases or biographies.

Award winning singer songwriter Josh Pyke says the government must act swiftly to force AI platforms to pay for using music to train machines. Picture: Supplied.

Award-winning artist Josh Pyke said his experience of using AI for music video ideas had been disappointing because “everything looks the same.”

“Creatives are always the first ones to use technology and embrace innovation,” he said.

“Synthesisers, electric guitars, social media, we love tools that help us create.

“But tech companies are using our work for commercial purposes, to monetise the data, and we want fair remuneration for the use of our songs to train AI.”

The report also called for streaming platforms to promote the songs created by humans over those generated by AI.

Missy Higgins fears AI will rob music of “what it means to be human.” Picture: Tajette O'Halloran / Supplied.

As Pyke said: “You have already got Jimmy Barnes; why would we need AI Jimmy?”

Jimmy Barnes and Missy Higgins said AI versions created by machines trained using songs composed by humans would lack “genuine emotion”, the magic ingredient of a hit song.

“It might be possible for Intelligence to be Artificial, but most music is driven by the heart, not the head, and I don’t think artificial feelings can stir the soul,” Higgins wrote in the report.

“If we start blurring the lines between genuine emotions and the machine-generated versions, then we risk losing a key part of what it means to be human.”

The AI and Music report comes as Australian artists have been hit by a flurry of blows to their sustainability over the past five years, including the collapsing festival market, diminishing income from streaming and algorithmic bias which floods playlists and airwaves with international releases over homegrown music.

Peter Garrett says the music industry is already under siege without losing more income for their creativity to AI platforms. Picture: Belinda Rolland

An overwhelming 97 per cent of the 4200 music creators who responded to the report’s survey demanded tech companies disclose when they use songs as training data, and for the government to introduce regulations so songwriters are given “adequate credit, consent and fair remuneration for any works being used in AI platforms.”

Midnight Oil frontman Peter Garrett said the “risks currently outweigh the opportunities with AI.”

“The music industry is already under siege, treated like a mendicant cottage industry, and we still don’t have effective action on things like much Australian work sinking under the Spotify algorithm,” he said.

“Without robust laws to ensure copyright holders are adequately remunerated, licenses applied and transparency around the actual processes used when a creator’s work is exploited, then

we‘re in deep trouble. No more speeches, papers, or promises – we need urgent government action now.”

The full report is via apraamcos.com.au/AIandMusic

Originally published as Aussie songwriters demand AI platforms pay for using their songs for training