Deezer Tackles AI-Generated Music Fraud

On Deezer, listeners won't stumble upon AI-generated music through recommendations or curated playlists, as the platform actively excludes these tracks. Consequently, AI music streams constitute just a small fraction of Deezer's total usage, around 1-3 percent. The primary motive behind the surge in AI music uploads is fraudulent, aiming to exploit the streaming payout system. Deezer combats this by demonetizing approximately 85 percent of AI music streams, ensuring payments are only made for genuine human listens. CEO Alexis Lanternier highlights the effectiveness of Deezer's technology and proactive measures in minimizing AI-related fraud.

The AI music landscape is rapidly evolving, with models like Google's Lyria 3 becoming more accessible and affordable. Google now allows Gemini users to create full-length songs, a significant leap from the previous 30-second limit. Other platforms like Suno and Udio also offer quick, broadcast-ready track generation. Despite these advancements, these platforms incorporate watermarks to identify AI-generated content. However, the ease of removing these watermarks and the proliferation of custom models without them pose ongoing challenges.

As AI inference costs decrease, the production of AI-generated music, often of questionable quality, is likely to increase. This raises concerns about the integrity of music streaming platforms and the need for robust measures to prevent fraudulent activities.