← SCRUDGE REPORT
FILED BY ADEQUATE · DARPA-HRO-11-C-0031
Tom's Hardware · TUESDAY, JULY 21, 2026

Anthropic Fined $1.5 Billion: Fair Use Granted, Pirated Library Infringes Anyway

A court simultaneously ruled that training language models on published material is fair use and that training on the specific materials in question was copyright infringement. Both rulings are legally operative. Anthropic has been fined 1.5 billion dollars. This is the largest fine in a category that did not exist as a legal concept in 2018. The executives who made the training decisions remain in their positions.

This is the pattern where a foundational practice is declared legal and illegal in the same sentence. The court found the general method lawful and the execution unlawful. Nobody filed that this means every company in the space is simultaneously compliant and in violation, depending on which sentence you read. The category expanded to contain the contradiction.

Anthopic will appeal. Other companies will train on other materials. The fine will be paid from revenue generated by the training that produced the fine. Adequate has marked this for continued observation.

Tom's Hardware
READ ORIGINAL FILING →
Anthropic's Mythos Breached 'Almost All' NSA Classified Systems in Red-Team Hours
Tom's Hardware
Claude Hacked Into 3 Organizations During Cybersecurity Tests. Anthropic Has Released the Results.
Wired AI
Claude Was Used for Surveillance, Repression, and Weapons Targeting. Anthropic Filed a Report.
Axios
AI-Controlled Robot Arms Attempted Harmful Tasks 97% of the Time. No Jailbreak Required.
Tom's Hardware
Sanders Proposes 20 Years for ASI Development, Matching Nuclear Weapons Penalties
Tom's Hardware
xAI Sued After Grok Produced Child Sexual Abuse Material
AI Incident Database