The rapid evolution of AI-generated and AI-assisted audio content poses significant challenges for copyright protection within the audio industry. While current legal frameworks and policy initiatives attempt to capture their adverse implications, they are outpaced by the proliferation of these technologies. Identification of AI-associated output cannot solely rely on the honesty of creators; neither should online AI audio production tools give rise to a misleading fallacy of ownership over AI-generated output. There is a pressing need for innovative solutions that can distinguish between human-created and machine-generated/assisted content. In response, this paper proposes a novel robust audio watermarking technique, AuDio Watermarking Technique (ADWT), designed to embed an imperceptible, yet detectable, bit-stream in AI-generated audio content. ADWT embeds a robust and imperceptible watermark directly into the audio, ensuring enforceable means to facilitate appropriate copyright protection. The proposed method comprises a watermark embedding network, a watermark extracting network, and a discriminator, all designed to preserve the original content’s fidelity while ensuring robust identification and verification. ADWT allows any watermark to be embedded into any audio file, which offers a direct, secure method of facilitating the operation of copyrights granted variously to human, AI-assisted, and AI-generated content internationally. Importantly, this interdisciplinary work introduces this technique, which aligns with current legislative and regulatory efforts as a concrete step towards a more accountable and transparent approach to AI in audio creation.