跳到主要內容

Anthropic版權訴訟評測:索尼華納提告,每首歌求償15萬美元 | Anthropic Lawsuit Review: Sony, Warner Sue Over Claude

By Kit 小克 | AI Tool Observer | 2026-09-06

🇹🇼 Anthropic版權訴訟評測:索尼華納提告,每首歌求償15萬美元

Anthropic版權訴訟在9月初正式開打,索尼音樂出版(Sony Music Publishing)與華納查普爾音樂(Warner Chappell Music)聯手在北加州聯邦法院提告,指控Anthropic為了訓練Claude,「大規模、蓄意」透過BT下載與爬蟲手法非法取得數萬首歌詞與詞曲版權,求償金額上看數十億美元,甚至直接點名創辦人Dario Amodei與Benjamin Mann為個人被告。

Anthropic版權訴訟在告什麼?

兩大出版商指控Anthropic為了訓練Claude,透過Library Genesis、Pirate Library Mirror等盜版資料庫,以及Musixmatch、LyricFind等歌詞網站,大量下載受版權保護的詞曲內容。控告書中點名的歌曲包括:

  • 〈Eye of the Tiger〉
  • 〈September〉
  • 〈Uptown Funk〉
  • 〈Hallelujah〉
  • 泰勒絲〈Paper Rings〉

訴狀用「brazen campaign(厚顏無恥的行動)」形容Anthropic的做法,稱這是史上規模最大、最露骨的智慧財產權竊取案之一。

求償金額有多驚人?

依美國著作權法,故意侵權的法定賠償上限是每件作品15萬美元,另外移除版權管理資訊(CMI)每件再罰2.5萬美元。以原告聲稱「數以萬計」侵權作品估算,總求償金額恐衝上數十億美元。更值得注意的是,這次訴訟不只告公司,還直接把Dario Amodei和Benjamin Mann列為個人被告,等於創辦人本人也要扛法律責任。

這跟先前的Bartz和解案有什麼不同?

Anthropic去年才在Bartz v. Anthropic一案中,以15億美元和解作家集體訴訟,創下AI版權賠償紀錄。這次索尼、華納把戰線拉到音樂出版,且原告直接主張「故意侵權」以爭取更高的法定賠償上限,顯示Anthropic版權訴訟戰場正從文字作品擴大到音樂產業,策略也比Bartz案更強硬。

對AI產業代表什麼?

Anthropic回應稱訓練生成式AI屬於「轉化性合理使用(transformative fair use)」,將全力抗辯。但兩大音樂版權龍頭聯手出手,加上直接告創辦人個人,等於向全產業放話:訓練資料來源不乾淨,創辦人也可能要自己扛責。對正在採購或微調LLM的團隊來說,這場官司值得持續追蹤——資料授權乾不乾淨,未來恐怕不再只是公司的事。

好不好用,試了才知道。


🇺🇸 Anthropic Lawsuit Review: Sony, Warner Sue Over Claude

The Anthropic lawsuit just escalated in a big way: Sony Music Publishing and Warner Chappell Music filed a joint federal complaint in Northern California, accusing Anthropic of a "brazen campaign" of torrenting and scraping tens of thousands of copyrighted songs to train Claude — and naming co-founders Dario Amodei and Benjamin Mann personally as defendants.

What Is the Anthropic Lawsuit About?

The complaint alleges Anthropic pulled copyrighted lyrics and compositions from pirate sources like Library Genesis and Pirate Library Mirror, plus lyrics sites Musixmatch and LyricFind, to train Claude. Named works include:

  • "Eye of the Tiger"
  • "September"
  • "Uptown Funk"
  • "Hallelujah"
  • Taylor Swift's "Paper Rings"

The publishers call it one of the largest and most blatant intellectual property thefts in history.

How Much Money Is at Stake?

Under U.S. copyright law, willful infringement carries statutory damages up to $150,000 per work, plus $25,000 per removed copyright management notice. With "thousands upon thousands" of alleged infringements, total exposure could run into the billions. Unusually, the suit also names Amodei and Mann as individual defendants, putting personal liability on the table alongside the company's.

How Does This Compare to the Bartz Settlement?

Anthropic already paid $1.5 billion last year to settle Bartz v. Anthropic, a class action from authors over book piracy — the largest copyright payout in AI history. This new case shifts the battlefield to music publishing and leans harder on "willful" infringement to unlock the higher statutory cap, signaling the Anthropic lawsuit front is expanding from text to music, with a tougher legal strategy than Bartz.

What Does This Mean for the AI Industry?

Anthropic maintains that training generative AI is "transformative fair use" and plans to fight the case. But two major publishers teaming up — and going after founders personally — sends a clear signal to every lab: dirty training data can now become a personal liability, not just a corporate one. If your team is licensing or fine-tuning on scraped data, this case is worth watching closely.

好不好用,試了才知道 / You won't know if it works until you try it.

Sources / 資料來源

常見問題 FAQ

Anthropic版權訴訟的原告是誰?

索尼音樂出版(Sony Music Publishing)與華納查普爾音樂(Warner Chappell Music),兩家聯手在北加州聯邦法院對Anthropic提告。

這場官司求償多少錢?

依美國著作權法,故意侵權每件作品法定賠償上限為15萬美元,加計數萬首歌曲後,總求償金額恐達數十億美元。

Anthropic的創辦人也會被告嗎?

會,訴訟直接點名共同創辦人Dario Amodei與Benjamin Mann為個人被告,而不只是告公司。

Anthropic之前有類似的版權官司嗎?

有,Anthropic去年才以15億美元和解Bartz v. Anthropic一案,是目前AI版權訴訟史上最大的和解金額。

Anthropic對這起訴訟的回應是什麼?

Anthropic表示訓練生成式AI屬於「轉化性合理使用」,將全力抗辯到底。

延伸閱讀 / Related Articles


AI 工具觀察站 — 每日精選 AI Agent 與工具趨勢
AI Tool Observer — Daily curated AI Agent & tool trends

留言

這個網誌中的熱門文章

Google Ironwood TPU v7 推理專用晶片解析:效能追平 NVIDIA、成本低 44%,AI 晶片戰爭正式開打 | Google Ironwood TPU v7 Explained: Matching NVIDIA Performance at 44% Lower Cost — The AI Chip War Heats Up

Claude Code 實測:AI 幫你寫程式到底行不行? | Claude Code Review: Can AI Really Code for You?

Cursor vs GitHub Copilot vs Claude Code:AI 程式助手大比拼 | AI Coding Assistants Compared: Cursor vs GitHub Copilot vs Claude Code