June

2 events

Editors' summary

A court separates training from sourcing, Claude enters classified work

On 23 June, Judge William Alsup held that training on books was fair use, while keeping a library assembled from pirate sites was a separate matter to answer for. The ruling drew attention for treating the training itself and the sourcing of the data as distinct questions.

On the 5th, Anthropic released Claude Gov for national security work — a version built for classified environments, able to handle material the standard models decline.

This block is written by the editors. It is kept separate from the sourced record below.

Record2 events

5
ProductAnthropic
Anthropic unveils Claude Gov models for national security
Anthropic released Claude Gov, a set of models offered only to US national security customers operating in classified environments, saying agencies at the highest level of US national security were already using them. Built on direct feedback from government customers, the models were tuned for handling classified material, understanding documents specific to intelligence and defense work, identifying languages and dialects, and interpreting cybersecurity data. Anthropic said they had gone through the same safety testing as its other Claude models. It marked a step-up in its government business following the Palantir and AWS partnership of November 2024.
23
PolicyAnthropic / William Alsup
Court rules AI training is fair use, but pirated library is not
In Bartz v. Anthropic, Judge William Alsup of the Northern District of California ruled on summary judgment that using books to train a large language model is transformative — "spectacularly so" — serving a purpose fundamentally different from the authors' purpose of being read, and that scanning lawfully acquired books for training was fair use. He held separately that assembling a central digital library from pirated copies was not fair use, and sent that question on toward trial. It was the first US court decision to address AI training data head-on, and it established that how material was acquired could matter more than what was learned from it — becoming a reference point for the copyright cases that followed.

↑ 2025

Anthropic unveils Claude Gov models for national security

Court rules AI training is fair use, but pirated library is not