From Audio Deepfake Detection to AI-Generated Music Detection -- A Pathway and Overview

This is a survey paper (0 stars, 0 velocity, arXiv preprint) that synthesizes existing deepfake detection literature and applies it to music detection—a well-trodden pathway rather than novel methodology. The work appears to be a literature review with conceptual connections between two established detection domains, not a new algorithm, dataset, or implementation. Zero GitHub adoption (0 stars, 4 forks likely from automatic mirrors) and no active development indicate this is an academic exercise without production intent or community validation. The techniques surveyed (spectral analysis, neural classifiers, artifact detection) are standard in audio ML and commoditized across multiple commercial and open-source projects (e.g., Authenticity.ai, Google's SynthID, various academic benchmarks). Frontier labs have little incentive to replicate: they either own the generation systems (reducing detection need in their context) or can integrate existing detection methods into their platforms. The survey's value is informational, not as a replicable system. Low defensibility due to: (1) no novel technical contribution, (2) no implementation/codebase, (3) re-application of known methods, (4) zero community adoption, (5) pure reference material. Frontier risk is low because this doesn't solve a proprietary problem they need to own; it's educational content they could cite or subsume into documentation.