Anthropic‘s $1.5 Billion Copyright Settlement Faces scrutiny, Highlights Disparities in Enforcement
SAN FRANCISCO – A proposed $1.5 billion settlement between AI company Anthropic and a class of authors alleging copyright infringement in the training of its large language models is facing potential roadblocks, as U.S. District judge William Alsup expressed concerns about the claims process and requested further details.The case, stemming from Anthropic’s use of datasets like LibGen and PiLiMi, underscores a basic imbalance in copyright enforcement – one where corporations face considerably lower penalties than individual citizens for similar infringements.
The lawsuit accused Anthropic of violating copyright by using illegally obtained books and other copyrighted material to train its AI models. While the settlement represents a record payout for a copyright class action, experts caution it’s narrowly focused, offering Anthropic release from past liability without establishing a broader licensing framework for future AI development. dave Hansen of the Authors Alliance blog noted the settlement “isn’t far-reaching,” unlike the Google Books Settlement which aimed to create a novel licensing scheme.
This disparity in enforcement is starkly illustrated by the case of Jammie Thomas, a single mother who in 2007 was initially found liable for $222,000 – later increased to $1.92 million on retrial – for sharing 24 songs on the peer-to-peer file-sharing service Kazaa. The judge in that case deemed the original damages “unprecedented and oppressive” and urged Congress to revise copyright law regarding P2P networks.
In contrast, Anthropic is offering to pay $1,500 per work allegedly infringed upon, a figure dramatically lower than the $80,000 per work Thomas faced. “This confirms once more that when it comes to copyright and its enforcement, there is one law for the rich corporations, and another law for the rest of us,” writes Glynn Moody on WalledCulture.org.
Judge Alsup has indicated he “felt misled” and requires additional details regarding the settlement’s claims process before approving the agreement.The outcome of this case will likely have important implications for the rapidly evolving landscape of AI and copyright law.
