Key Takeaways

  • Josh Fawaz’s chart-topping "Like a Prayer" rework is intensifying scrutiny of AI use in commercial music production.
  • ARIA plans to adopt new IFPI and RIAA global AI labelling standards that will require explicit disclosure of synthetic generation.
  • Broadcasters and rights organizations face growing pressure as AI-generated tracks compete for royalties and airplay.

In a matter of months, previously low-profile Australian producer Josh Fawaz watched his rework of Madonna’s "Like a Prayer" climb to the top of the National Radio Airplay chart and drive substantial streaming numbers. Yet the track's ascent also triggered a sharp reaction across Australia’s music and broadcast sectors. Claims that generative AI may have powered the vocals or production have amplified long-standing concerns about attribution, royalties, and transparency.

Fawaz denies that the vocals are fully synthetic, stating he uses AI strictly as a production tool. Still, the discussion arrives as the industry is already under pressure to adapt. Technology companies are pushing for expanded dataset-use rights, broadcasters are experimenting with AI voice hosts, and global standards bodies are beginning to set formal rules about AI-produced audio.

The controversy surrounding the track landed during the same week global music bodies IFPI and RIAA agreed to a new AI labelling system distinguishing AI-generated from AI-assisted recordings. ARIA publicly backed that standard and signaled it will adopt the framework for Australian releases. Future tracks utilizing generative production tools would require explicit disclosure once implemented, addressing the transparency gap that critics currently find troubling.

Some observers, including a senior research fellow at RMIT's school of media and communication, argue the track carries hallmarks of generative platforms, pointing to heavy vocal compression. Others note that commercial radio embraced the song without public inquiry into its creation. This lack of scrutiny contrasts with growing expectations across other media to disclose AI usage. As of July, Australian radio hosts must explicitly identify AI voices on air, though the rule currently exempts music.

Platforms such as ElevenLabs, Suno, and AIVA are accelerating new forms of audio production. Analysts tracking this shift, including researchers from the MIT Media Lab, have examined how synthetic voice tools enable rapid content pipelines, while Deloitte has reported that rights management systems lag behind accelerated automation in creative industries. Broadcasters and labels looking to shape responsible use policies are increasingly referencing structured guidance like the 2023 NIST AI Risk Management Framework.

Against this backdrop, the commercial impact of AI-assisted or AI-generated tracks is driving debate across the sector. One commercial producer went public with concerns that AI tracks collect royalties and directly compete with human-created music, pointing to production artifacts that experienced listeners can identify, such as odd vocal textures and low-fidelity stems. The resulting conversation about royalty distribution remains unresolved.

APRA AMCOS, which handles rights payments for musicians, confirmed that the original rights holders for the underlying composition of "Like a Prayer" will continue to receive performance royalties for the musical work. While royalty allocation for compositions is settled, the industry faces a harder question regarding royalties tied to AI-generated performances. Legal experts are questioning whether traditional contracts apply in cases where a text prompt, rather than a human performer, generates the vocal line.

Meanwhile, the Media, Entertainment and Arts Alliance is advocating for mandatory watermarking of AI content and for equitable remuneration to be added to Australian copyright law. Multiple analysts, including teams at McKinsey, have noted that creative labor models undergo rapid restructuring whenever automation becomes widespread. As thousands of tracks produced partially or entirely by AI compete in the same royalty pool as human artists, radio rotation and playlisting algorithms are likely to shift.

The current debate reflects the culmination of a decade of streaming-driven fragmentation. Researchers suggest that modern listening habits have conditioned audiences to be less attentive to the exact origin of a track. However, as generative audio platforms reach mainstream visibility and AI-associated tracks top national airplay charts, the lack of clear attribution escalates from a niche production debate into a tangible business issue.

Broadcasters face an immediate operational challenge. With AI voice hosts already being piloted, stations must determine how to tag, catalogue, and report synthetic content at scale. Media companies are studying general risk principles from bodies like NIST to develop workable policies for complex entertainment use cases.

Adoption of the new IFPI and RIAA labels is expected to bring clarity to the market. Designating a track as AI-assisted or AI-generated provides institutions the ability to make informed programming choices. The immediate challenge lies in the transitional period, during which neither creators nor distributors are formally required to disclose generative AI usage.

While Fawaz defends his artistic process, commercial radio is confronting a rapidly shifting landscape. The widespread charting success of tracks with contested origins demonstrates that generative AI is forcing music’s legal and economic infrastructure to evolve. The industry must now determine how quickly that evolution will happen and how rights frameworks will adapt once clearer standards take hold.