Judge orders Meta to name who authorized pirated training-data downloads
What happened
A report on October 21 says Meta is trying to avoid disclosing which executives approved torrenting pirated material for AI training, and pins the downloads on employees. Meta says early Llama 1 and 2 training data came through informal processes, with approval arising from collaborative discussion and no single decision-maker. Judge Hixon ordered Meta to say more about who took part in and authorized the decisions, noting the request covers all decision-makers, not only those who made formal decisions. Meta must also explain what happened to pirated works downloaded but not used in training.
Written by AI from the coverage · updated 1 hour ago
Coverage
Follow the reports to see the story from different sides.
- Ars Technica · AIMeta tries to scapegoat staff who downloaded torrents to spare execs
Meta 试图避免披露批准 AI 训练数据种子下载的高管，Hixon 要求其进一步交代决策及授权人员。Meta 称 Llama 1 和 2 早期训练的数据获取采用非正式流程，且批准来自协作讨论，没有单一决策者；Hixon 指出，相关请求涵盖所有决策者，并不限于正式决定。Meta 还须回答未用于 AI 训练的盗版作品最终如何处理。
Heat over time
Not enough continuous observations to draw a trend yet.