AI Training Data: Courts Say Shredding Rare Books Is Legal
A story about AI companies physically destroying rare books to scan them for training data surfaced and generated a sharp thread. The key line from comments: 'You can reprint a bestseller. You can't replace the last three copies of an 18th-century botanical text once someone shreds them for training data. And the judge said it's legal.' References to Fahrenheit 451 and Vernor Vinge's 'shred and scan' factory showed up, which is the thread's way of saying this feels like something from a dystopia.
The legal clearance is the important part. Courts are not yet drawing a line between digitizing a book and destroying the physical artifact in the process. This is not a copyright question, it is a cultural preservation question, and the AI industry currently has no answer for it.
The counterpoint in the thread was asking which specific rare titles were destroyed, suggesting some skepticism about whether the alarming framing matched the actual scale of destruction. But even if the current volume is small, the legal precedent is established.
So what?
Founders building AI training pipelines or data acquisition businesses need to know that physical destruction of source materials for scanning is legally permissible right now. The reputational risk, however, is real and growing. If you are acquiring archival or rare materials, document the physical condition and disposition of every item.