Rapid
progress in AI technology, constituting an
AI boom, was brought to public attention in the early 2020s by text-to-image models such as
DALL-E,
Midjourney, and
Stable Diffusion, which were able to generate complex images that convincingly resembled human-made artworks. The proliferation of such image generation algorithms coincided with the release of
GPT-3 and development of
GPT-4, advanced
large language models which produce highly convincing text. These transformer-based models designed to create new content from prompts are collectively called
generative artificial intelligence, and they require vast sets of
training data. This data often consists of text, images, and other media
scraped from the web, prompting concerns that the AI products may violate
intellectual property rights.
Suno AI and
Udio, two AI startups whose products generate music recordings following user-submitted prompts, were sued in 2024 by
Sony Music,
Warner Music Group, and
Universal Music Group, who alleged that the companies used copyrighted recordings in their training data without authorization. In December 2024, the UK government announced a consultation on copyright and AI, outlining a preferred approach that would see the introduction of a
data mining copyright exception with a rights reservation package for rights holders. In the months following the announcement of the consultation, a number of prominent musicians warned of the threat it posed to musicians, including
Paul McCartney and
Elton John. ==Content==