Kylie Minogue and Nick Cave are among the Australian artists whose work has already been used to train artificial intelligence models without compensation, according to the International Confederation of Music Publishers, as the global body warns policymakers against weakening copyright laws under pressure from tech companies.
The Bottom Line
- The International Confederation of Music Publishers reports that AI companies have already trained models on music by Australian artists like Kylie Minogue and Nick Cave without paying creators.
- Australia stands out as the sole jurisdiction among 62 countries considering legal changes to allow text and data mining exceptions for commercial AI training.
- A parliamentary inquiry’s leaders recently recommended examining current copyright settings to ensure creators maintain control and receive payment, though an independent senator has raised transparency concerns.
Global Music Industry Warns Australia Over Copyright Exceptions
The International Confederation of Music Publishers, which represents major companies like Sony, Warner, and Universal alongside thousands of independent publishers across six continents, stated that AI companies are pushing policymakers to legitimize unauthorized data use. John Phelan, the ICMP’s managing director, told The Guardian that technology firms are attempting to strong-arm Australian lawmakers into lowering copyright protections by threatening to withhold investment.
Anthony Albanese remarked in July that utilizing artists’ work to build and train artificial intelligence without their control or financial compensation constitutes theft. Despite this stance, organizations across the music, film, media, and news sectors voiced concerns that global technology companies have lobbied the government behind closed doors to weaken local copyright frameworks.
Parliamentary Committee Re-examines Local Copyright Settings
Government and opposition members of a parliamentary committee focused on artificial intelligence described the current environment as a stalemate this week. The committee noted that while AI models already train on copyrighted material in the United States without compensating creators, technology firms have withheld training frontier models locally due to existing Australian copyright rules.
In a joint letter submitted to the prime minister, Labor inquiry chair Jo Briskey and Liberal deputy chair Senator Jonathon Duniam recommended that the government examine current copyright settings while guaranteeing that creators retain control and receive compensation. Independent senator David Pocock has raised concerns that the committee adopted a predetermined position on copyright after staffer edits appeared in the correspondence, a claim that Briskey rejected.
Evidence of Unauthorized Training on Australian Catalogues
Technology companies have maintained that their models are not trained on Australian content because of strict local copyright rules. However, evidence provided exclusively to The Guardian by the ICMP demonstrated that Google’s Gemini successfully reproduced lyrics from Australian musicians including Kylie Minogue, G-Flip, and Nick Cave.
Phelan also highlighted a dataset reported by The Atlantic in June containing more than twelve million songs harvested from YouTube videos, which includes tracks from Minogue, Cave, and the Hilltop Hoods. Google was approached for comment.
International Resistance to Statutory Loopholes
The ICMP has monitored global efforts by AI companies to alter intellectual property laws. Phelan stated that out of sixty-two countries that have considered legal adjustments, Australia appears to be the only nation entertaining industry demands for commercial text and data mining exceptions.
“There isn’t a single country among the other 61 … which has introduced a new text and data mining exception for commercial purposes, or reduce the idea of exclusive rights and copyright,” Phelan said, adding that the international creative community is closely watching how Australia handles the issue.
Representing Anthropic, special envoy Jeffrey Bleich told the AI committee that his company is not attempting to dictate terms to the government, but rather explaining the implications of training “with respect to our capacity to operate here.” “You can’t license the entire internet. Billions and billions of information exists there which is available to be viewed and studied, and that’s essentially what training is,” Bleich said. Meanwhile, OpenAI’s chief of economic policy, Adam Cohen, pointed to US fair use principles and noted that his firm has established commercial deals with outlets such as News Corp and The Guardian despite broader legal arguments regarding public data.
