Genshin Impact Privacy Policy quietly changed after revealing voice data used for AI training

Genshin Impact's Privacy Policy indicated players' voice comms data may be used to train AI models

Genshin Impact's privacy policy featured a disclaimer indicating users' voices were being used to train AI. This privacy policy, last published this January, has since been quietly updated to remove this clause. Spotted by user Cevian on Bluesky, section 1 subsection xviii of the Policy, 'what data we collect and process' prior to the clauses' removal stated: "Where applicable, our services may offer voice-based chat functions that allow users to communicate with others in real time using voice. If you choose to use such features, we may process voice communication data for the purposes described in section 2 and section 13 of this Privacy Policy, including providing communication services, maintaining service security and stability, content moderation, and complying with applicable laws and regulations." This subsection has since been removed entirely Section 2 of the Privacy Policy, titled 'why we collect and process your personal data' states, among many other uses: "Like many AI-powered experiences, we may use your user-generated input such as chat data to train and improve the model that we use to provide our Services. You can opt-out of model training anytime in your Service-related settings. Opting out will not affect your existing gameplay." It is worth noting that this was just once clause of Section 2, and other clauses including providing customer support and preventing malicious content to "ensure a harmonious and healthy online environment" are included. This same policy states that users must opt out of model training through service-related settings. MiHoYo has been open about its embrace of generative AI in the past. It released a teaser for an upcoming Unreal Engine 5 MMO project that used generative AI late last year. This game, while largely still a mystery, will feature: "Artificial Intelligence Stimulated Eco System", All-Terrain Battle AI, and "Intellectual NPC". Eurogamer has contacted MiHoYo for comment on exactly what users' voice communication data is being used for, and why the subsection related to the gathering of such data was removed from the Privacy Policy.

A Privacy Policy Change has Genshin Impact Players Concerned Their Voice Chat Data was Used to Train AI Models, Unless They Opted Out

With Generative AI (GenAI) remaining the most controversial topic in the video game industry, any mention of GenAI or AI in general is making the hair on players' arms stand up. The latest occurrence of this comes from Genshin Impact players looking a little more closely at Hoyoverse's recent updates to the popular free-to-play game's Privacy Policy, which appears to have indicated that unless players opt-out, Hoyoverse will collect players' in-game voice chat data to train an internal AI model. Initially spotted on a now-removed Reddit post, Bluesky user Cevian shared a change to Genshin Impact's privacy policy that was made as recently as January 14, 2026. The change specifically referred to a clause from Section 1, subsection XVIII, being removed, which read, "Where applicable, our services may offer voice-based chat functions that allow users to communicate with others in real time using voice. If you choose to use such features, we may process voice communication data for the purposes described in Section 2 and Section 13 of this Privacy Policy, including providing communication services, maintaining service security and stability, content moderation, and complying with applicable laws and regulations." Now, with the above section removed, Section 2 of the Privacy Policy, titled "Why We Collect and Process Your Personal Data," simply reads: "Like many AI-powered experiences, we may use your user-generated input such as chat data to train and improve the model that we use to provide our Services. You can opt-out of model training anytime in your Service-related settings. Opting out will not affect your existing gameplay." It's worth noting that the now-removed subsection XVIII statement is referring to Hoyoverse using the data it collects for things like improving its customer support and its moderation efforts. Essentially, there's no indication that Hoyoverse is using this data to train an internal GenAI model, which appears to be part of the concern players had when this policy was first spotted. We've reached out to Hoyoverse for a comment on the situation, and despite the fact that it doesn't appear to be the case that Hoyoverse would be using this data for anything unsavory, the concern players have seems to mostly come from a feature like this being on by default, with users having to manually opt out. That's what caused players to have issues with Microsoft's Copilot Gaming AI, which added an 'on-by-default' feature that initially appeared to be recording your gameplay to train Copilot. That wasn't exactly the case, but the fact that it was on automatically, without players being sufficiently asked about it in the first place, is what caused the uproar around it. And it's worth betting that people would care more about their voices being recorded without their knowledge or proper consent, over some of their gameplay footage. Hopefully, we'll have some word from Hoyoverse soon to properly explain the situation.