Recommended Articles
Legal & Policy

Shanghai Court Rules Against AI Company Over Genshin Impact Voice Cloning

By Angeled · October 28, 2026 · Legal & Policy
Shanghai Court Rules Against AI Company Over Genshin Impact Voice Cloning
Shanghai Court Rules Against AI Company Over Genshin Impact Voice Cloning

A Shanghai court has found copyright infringement and unfair competition in a case involving an artificial intelligence service that imitated video-game character voices and used game imagery from Genshin Impact without permission. The ruling, reported on 28 October 2026, is one of the clearest judicial statements yet on where AI voice services cross the line from imitation into infringement — and it arrives as voice cloning becomes one of the fastest-growing, and least legally settled, corners of the AI economy.

Public reporting of the decision is so far limited to a specialist legal summary, which does not name the defendant company or quantify any damages. What is established is the core holding: a service that imitates character voices and uses game imagery without permission can be liable on two distinct grounds — copyright infringement and unfair competition. That dual basis matters, because it gives rights holders two routes to relief and makes the ruling harder to design around than a single-issue copyright decision would be.

Why character voices are legally special

A game character's voice is not a single legal object. It can implicate several overlapping rights at once. The recorded performance may be protected as a sound recording or as part of an audiovisual work. The character itself — its design, name, dialogue and visual identity — may be protected by copyright. The voice, particularly where it is distinctive and associated with a known character, may also be protected against misappropriation of likeness or personality, and its unauthorised commercial use can mislead consumers into believing a product is endorsed by or affiliated with the game.

AI voice cloning complicates each of these layers. A cloning service typically does not copy a recording directly; it trains a model on recordings and then generates new audio that sounds like the character saying new words. Rights holders argue that this still appropriates the value of the performance and the character, and that offering such a service commercially — especially where it is marketed with the game's imagery — trades on the game's reputation. Defendants tend to argue that a voice, as such, is not copyrightable, that training is a transformative use, or that the output is original expression. The Shanghai court's ruling rejects the practical version of that defence for a commercial service built around recognisable characters: imitation plus unauthorised imagery, offered to the public, is infringement and unfair competition.

How this fits the global voice-cloning crackdown

The Shanghai decision is part of a worldwide pattern, not an outlier. Courts and regulators in multiple jurisdictions have been moving against unauthorised AI replicas of real and fictional voices. In the United States, state laws — most prominently Tennessee's ELVIS Act — now prohibit unauthorised AI replicas of a person's voice and likeness, and performers' unions have made voice protection a central bargaining demand. In the European Union, personality and related rights, together with copyright and unfair competition law, are being tested against similar services. China's own legal framework has been developing rapidly on this front, with courts previously recognising protection for AI-generated content and for voice rights in other contexts.

The distinction founders must understand

The ruling does not outlaw voice AI. It draws a line between voice technology in general and a service that imitates identifiable, protected characters and markets itself with their imagery. That distinction is the one founders need to internalise. A voice product built on licensed voices, on original characters, on consented voice talent, or on clearly generic voices sits on very different legal ground from a product whose selling point is "sound exactly like this famous character."

The compliance implications follow directly. Before launching a voice product, founders should be able to answer four questions: whose voice is being imitated, and is it licensed; whose imagery or branding is used in marketing, and is that licensed; is the output likely to confuse consumers about affiliation or endorsement; and what happens when a user prompts the system to imitate a protected character? If any of those answers is "we haven't checked," the Shanghai ruling is a warning that a court may not accept technical novelty as an excuse.

Why it matters for founders and investors

For founders in voice AI, audio, gaming and creator tools, this ruling raises the cost of the "clone first, ask later" strategy. Licensing is becoming a competitive advantage rather than a bureaucratic cost: companies that secure voice licences, build original character libraries, or partner with rights holders can market their products without the overhang of an infringement suit. Investors should treat unlicensed character imitation as a legal liability, not a growth hack, and should ask voice startups directly about their training data sources, their licensing position and their guardrails against imitating protected characters on user prompts.

Angeled.com connects founders with investors who back AI products built on licences, not lawsuits.

Source: mlex.com
MLex, "Shanghai court rules against AI company over Genshin Impact voice cloning," 28 October 2026

Read Next