Хөгжмийн салбарын томоохон төлөөлөгчид хиймэл оюун ухааны технологийн компани нь зохиогчийн эрхээр хамгаалагдсан дууны үгсийг зөвшөөрөлгүй ашигласан хэмээн буруутгаж байна.
Хойд Калифорнийн холбооны шүүхэд баасан гарагт гаргасан нэхэмжлэлд Anthropic компани болон түүнийг үүсгэн байгуулагчид болох Дарио Амодей, Бенжамин Манн нарыг оюуны өмчийг бүдүүлгээр зөрчсөн гэж үзжээ. Нэхэмжлэгч тал зөрчил тус бүрд 150,000 ам.долларын нөхөн төлбөр шаардаж байгаа бөгөөд энэ нь нийлбэр дүнгээрээ тэрбум ам.долларт хүрч болзошгүй юм.
Anthropic-ийн эсрэг хөгжмийн салбарынхны гаргаж буй анхны тохиолдол энэ биш бөгөөд өмнө нь Universal Music Group, BMG, Round Hill Music зэрэг компаниуд шүүхэд хандаж байсан. Энэ удаагийн нэхэмжлэлд дурдсанаар, тус компани нь MusixMatch, LyricFind зэрэг эрх бүхий сангаас дууны үгсийг хуулбарласан төдийгүй, тэдгээр нь Claude чатбот дээр шууд гарч ирдэг байна.
Нэхэмжлэгчид мөн Anthropic-ийн өмнөх хуулийн маргаантай холбоотой баримтуудыг ашиглан тэднийг системтэйгээр зохиогчийн эрх зөрчдөг хэмээн буруутгаж байна. Өнгөрсөн онд Anthropic нь ном зохиогчидтой холбоотой зохиогчийн эрхийн маргааныг 1.5 тэрбум ам.доллароор эвлэрэн шийдвэрлэж, зөвшөөрөлгүй ашигласан мэдээллийн сангаа устгахаар тохиролцсон байдаг. Одоогоор Anthropic компани энэ асуудлаар ямар нэгэн тайлбар өгөөгүй байна.
Дэлгэрэнгүйг эх сурвалжаас харах
↓Эх сурвалжийг нээх ↓
In what feels a bit like a flashback to the days just after the release of ChatGPT, music publishers are using very harsh language in a new lawsuit against Anthropic, alleging “one of the largest and most blatant ongoing thefts of intellectual property in history,” per Axios.
At first glance—and honestly after a few glances—it’s a bit of oddly timed legal saber-rattling, essentially all about song lyrics, and it seems calibrated to result in a drawn-out legal fight (or a lovely, big settlement for the plaintiffs).
The suit was filed late Friday night in Northern California federal court. It names Anthropic the company, as well as co-founders Dario Amodei and Benjamin Mann. The monetary damages sought—$150,000 per alleged infringement—potentially add up to billions.
You might be saying “Hey, aren’t all sorts of music publishers already suing Anthropic?” And yes they most certainly are, big and small.Universal Music Group, Concord, and ABKCO sued back in 2023; BMG sued in March; and the indie publisher Round Hill Music sued earlier this month.
The gambit with this latest suit appears to be a bifurcated approach. As laid out in the industry-friendly publication Music Business Worldwide, the suit partly leans on documents released thanks to an earlier lawsuit: Bartz v. Anthropic. That, you may recall, was the book author lawsuit that was settled last year with Anthropic agreeing to pay $1.5 billion. The company had allegedly pirated thousands of books by torrenting the Library Genesis (LibGen) and Pirate Library Mirror (PiLiMi) collections. In the end, Anthropic agreed to destroy the pirated copies it had used or planned to use for training data.
At the risk of editorializing in a legal matter without being a lawyer myself, unless there were lyric books in those allegedly pirated collections, this section of the lawsuit seems to be there mainly to establish a pattern of conduct that helps the plaintiffs’ case. Torrenting e-books, after all, is kind of a funny and roundabout way to get your hands on song lyrics.
But the lawsuit also alleges that Anthropic violated the plaintiffs’ copyrights by scraping authorized lyric repositories like MusixMatch and LyricFind. And lyrics can allegedly be coaxed out of Claude verbatim. As examples, the suit cites:
“[…]such beloved songs as ‘Ain’t No Mountain High Enough,’ ‘All I Want for Christmas is You,’ ‘Eye of the Tiger,’ ‘Here Comes Santa Claus,’ and ‘Paper Rings.’”
This is a more familiar, if still legally fuzzy, type of allegation, a bit like the famous New York Times lawsuit against OpenAI, Microsoft, and Perplexity. That suit bypasses any sort of fair use rationale when it comes to training data, and points instead to the idea that you can (allegedly) get ChatGPT to reproduce verbatim and substantive sections of paywalled New York Times articles. The legal thinking goes that this harms the New York Times, a fount of information, by potentially letting customers get that same information from a chatbot. That case is still unresolved.
Gizmodo reached out to Anthropic for a statement but did not receive a reply.
