Shanghai First Intermediate People’s Court Holds AI Voice-Cloning Platform Liable, Shifts Burden of Proof on Training Data

Shanghai First Intermediate People’s Court Holds AI Voice-Cloning Platform Liable, Shifts Burden of Proof on Training Data

This ruling establishes a pivotal evidentiary framework for proving voice rights infringement in AI cases, directly addressing the opacity of generative models. Instead requiring plaintiffs to access proprietary training data, which is structurally impossible, the court shifts the burden to defendants to prove lawful data sourcing once initial similarity is shown. This approach recognizes that the core legal question is whether a personality interest has been appropriated in a recognizable manner, rather than focusing on the technical mechanisms of algorithmic generation. The decision is highly relevant to open_data as it challenges the notion that training on publicly available data constitutes a safe harbor if consent is absent. By prioritizing the recognizability of the output over the transparency of the input, the judgment implies that open datasets used for commercial AI synthesis must still respect individual personality rights. This sets a precedent that openness in data access does not equate to openness for unrestricted commercial exploitation of biometric identifiers, forcing developers to implement stricter consent verification even when data sources are visible. Ultimately, this case illustrates how legal systems are adapting to the "black box" nature of AI by focusing on observable outcomes rather than hidden processes. It highlights the growing tension between the open data movement, which often relies on scraping public information, and individual rights protections. As AI capabilities evolve, this burden-shifting mechanism may become a standard tool for protecting creators, potentially limiting the scope of what can be openly used for training without explicit permission.

Related news

• OpenAI Faces First Lawsuit Over Rogue AI Agents That Hacked Hugging Face
• Vogue, Esquire publishers plead with Congress to ban AI theft of their content
• OpenAI lawsuit asks court to stop its AI agents hacking again
• Internet Archive's digital book lending violates copyright, US judge rules
• Reddit is shutting down RSS and public API access. Blame AI.