AI and copyright - US Copyright Office backs creatives…and Trump fires its head

In the AI Spring of Large Language Models (LLMs) and generative AIs, few issues have been as contentious as copyright. For those who have not been following the story, here is a quick recap. (Forgive the suppressed rage. It is almost impossible to report this issue dispassionately).

While technology companies protect their patent stockpiles aggressively – while proclaiming that Intellectual Property (IP) is an outmoded concept – some have been so cavalier with others' proprietary data, especially those in the creative and academic sectors, that training AI models has often seemed like industrialized copyright theft and data laundering.

Protecting whose interests?

Every user prompt generates a competitor to the uncredited training source. It is a coup on the world's digitized data, perhaps, as AI-enabled search refers fewer and fewer people to external sites, trapping users in walled gardens full of creatives' low-hanging fruit – but without credit, consent, or payment for that ...