Apple Sued For Training AI Model With Steeled Book

JAKARTA Apple harus menghadapi gugatan class action yang diajukan oleh sejumlah penulis buku ke pengadilan federal California Utara. Gugatan ini baru diajukan pada Jumat, 5 September lalu.

In the lawsuit, Apple is accused of using copyrighted books to train their Artificial Intelligence (AI) model. However, what Apple uses is not an ordinary book that is sold on the market, but data in the form of illegal.

Dua orang yang mengajukan gugatan, yakni penulis kirim Hendrix dan Jennifer Robertson, mengklaim Apple menggunakan setelah data bajakan yang dikenal sebagai Books3. Data ini digunakan untuk melatikan model bahasa OpenELM-nya.

These allegations were made based on the information Apple provided in its paper on OpenELM. In the paper, Apple referred to RedPajama as one of the datasets they used.

RedPajama itself uses Books3, data containing a collection of pirated books.

For this action, the authors demanded financial compensation and asked the court to order Apple to destroy the AI model using their work.

This isn't the first time a lawsuit over the use of books to train AI models has emerged. Previously, Anthropic was sued for using ordinary books to train AI, but the judge ruled that the company's actions were 'natural use'.

Anthropic was only asked to pay a fine. Meta is also one of the companies that won in a similar lawsuit.

Like Anthropic, Meta's actions in using copyright books to train AI models are considered a reasonable category.