Home · Health · Sep 18 archive

AI Copyright Fight Reveals 'Doom Loop' Worries

Developing

Health Desk

In Short: Tech companies feared AI could make newspapers obsolete, creating a 'doom loop' of copyright theft.

Tech companies have been grappling with the potential consequences of using copyrighted news articles to train AI systems, according to legal documents released by a U.S. court. The documents, part of an ongoing legal dispute between U.S. newspapers and AI companies like OpenAI and Microsoft, reveal concerns about a 'doom loop' where AI could render newspapers redundant.

The i Paper reported that the tech companies argued training AI on copyrighted news articles was 'fair use,' but Microsoft executives admitted this could undermine the concept of fair use. One Microsoft document cited in its legal brief acknowledged that the race to fuel AIs with more content could drive news sites out of business, starving AIs of fresh content.

The documents also quote a Microsoft executive describing the AI training process as 'an astonishing theft of unprecedented proportions,' noting that AIs are 'hoovering up' the work of millions of people. Another internal Microsoft document referred to the situation as a 'doom loop' that would hurt AI model performance and the entire web.

The legal fight highlights the tension between the need for vast amounts of data to train AI systems and the rights of content creators. The i Paper reported that the documents allege the person involved called it perhaps the 'largest theft of labor in human history,' underscoring the scale of the issue.

The unsealed documents, which are partly redacted, provide extensive quotes from the tech companies' internal conversations, shedding light on their concerns and strategies. This legal spat is expected to set important precedents for how AI uses copyrighted material in the future.

What's still developing

Sources