Google won’t just admit it’s feeding YouTube creators to its music AI
Google won’t just admit it’s feeding YouTube creators to its music AI
谷歌不愿承认正在利用 YouTube 创作者的内容训练其音乐 AI
If you’ve uploaded a song to YouTube, Google almost certainly considers your video fair game for training its Lyria music AI, it just won’t admit it right now. 如果你曾向 YouTube 上传过歌曲，谷歌几乎肯定会将你的视频视为训练其 Lyria 音乐 AI 的“公平游戏”素材，只是目前它还不愿承认这一点。
A group of independent musicians is suing Google, claiming that it illegally used songs they uploaded to YouTube to train its Lyria 3 model. Google has filed a motion to dismiss the case, saying: 一群独立音乐人正在起诉谷歌，声称该公司非法使用了他们上传到 YouTube 的歌曲来训练其 Lyria 3 模型。谷歌已提交动议要求驳回此案，并表示：
Their lawsuit is based on the unsupported hypothesis that Google trained on their specific works. Even accepting their untested allegations as fact, the Complaint cannot stand. Plaintiffs each granted YouTube, and Google — which provides the service—a broad license to use the uploaded content. That license, present in YouTube’s Terms of Service, authorized the conduct alleged in the Complaint. “他们的诉讼基于一种未经证实的假设，即谷歌使用了他们的特定作品进行训练。即使接受他们未经证实的指控作为事实，该诉状也无法成立。原告各自授予了 YouTube 以及提供该服务的谷歌广泛的许可，以使用上传的内容。该许可存在于 YouTube 的服务条款中，授权了诉状中所述的行为。”
This is standard hedging for a legal filing: “you can’t prove we did it, and even if we did, we’re allowed to.” When asked directly if Google was using YouTube videos to train its Lyria 3 AI music model, the company declined to comment. But based on past public statements, it seems safe to assume the answer is yes. 这是法律文件中标准的避险措辞：“你们无法证明我们做了，即便我们做了，我们也是被允许的。”当被直接问及谷歌是否正在使用 YouTube 视频来训练其 Lyria 3 AI 音乐模型时，该公司拒绝置评。但根据以往的公开声明，可以肯定答案是肯定的。
In an interview from April 2024 with Bloomberg, YouTube CEO Neal Mohan said that “some portion” of YouTube videos may be used internally to train models like Gemini. Later that year, a blog post about creator tools confirmed as much, saying “we use content uploaded to YouTube to improve the product experience for creators and viewers across YouTube and Google, including through machine learning and AI applications.” 在 2024 年 4 月接受彭博社采访时，YouTube 首席执行官 Neal Mohan 表示，“一部分”YouTube 视频可能会在内部用于训练 Gemini 等模型。同年晚些时候，一篇关于创作者工具的博文证实了这一点，称“我们使用上传到 YouTube 的内容来改善 YouTube 和谷歌上创作者与观众的产品体验，包括通过机器学习和 AI 应用。”
The company even confirmed to CNBC that it was using YouTube uploads to train Gemini and Veo. What Google hasn’t done is specifically confirm that it’s also using YouTube uploads to train Lyria. 该公司甚至向 CNBC 证实，它正在使用 YouTube 上传的内容来训练 Gemini 和 Veo。但谷歌尚未明确确认它也正在使用 YouTube 上传的内容来训练 Lyria。
However, in its motion to dismiss, Google says that by uploading content directly to YouTube, the plaintiffs agreed to the terms of service, which grants the company the right to “reproduce, distribute, [and] prepare derivative works,” based on the upload. 然而，在驳回动议中，谷歌表示，通过直接向 YouTube 上传内容，原告同意了服务条款，该条款授予了公司基于上传内容“复制、分发和制作衍生作品”的权利。
With all of that considered, it might seem odd that Google won’t just admit the obvious. But right now, the company has little to gain by going on record. And with litigation pending, maintaining plausible deniability is a calculated move. 综上所述，谷歌不愿承认这一显而易见的事实似乎有些奇怪。但就目前而言，该公司公开表态几乎没有任何好处。在诉讼悬而未决的情况下，保持“合理推诿”（plausible deniability）是一种经过深思熟虑的策略。