Paul McCartney, Elton John and other creatives demand AI comes clean on scraping

Who gets to complain about AI training?

• Some argue famous musicians are technically uninformed “weavers” resisting new tools, so their objections should carry little weight.
• Others counter that being directly economically affected makes them more legitimate stakeholders, not less.
• There’s pushback against framing rich artists as automatically unsympathetic, noting that distrust of big tech is at least as strong as resentment of celebrity wealth.

AI as tool vs exploitation of prior work

• One camp sees generative AI like drum machines or DAWs: a higher‑level tool that won’t kill human art but add new forms.
• Opponents say that analogy fails because AI models wouldn’t exist without massive ingestion of others’ work, often used to mimic artists or “make them say/do things” they never did.
• A recurring analogy: this isn’t “icemen vs refrigeration,” it’s “stealing the icemen’s ice to power the fridge.”

Copyright, consent, and platforms

• Several commenters want strict proof of consent for all training data, plus explicit opt‑in (not buried opt‑out) from platforms like YouTube or SoundCloud.
• Others note platforms may already have broad licenses that allow sublicensing for AI training, though critics question whether such consent was ever “informed.”
• There’s comparison to music sampling: courts forced clearance and royalties; some expect a similar outcome for training data.

Scraping vs piracy and “data laundering”

• Some distinguish legal web scraping from “pirating” whole copyright libraries or book torrents to train models.
• The metaphor of “data laundering” appears: raw copyrighted content goes in, an opaque model comes out, and companies claim it’s no longer traceable.
• Commenters emphasize many people posted under old terms that never contemplated AI use, so current reuse may be ethically or legally dubious.

Law, enforcement, and geopolitics

• One side fears that strict consent rules would handicap the West versus countries that ignore them.
• Others reject “ends justify the means” reasoning, arguing technological advantage doesn’t excuse mass uncompensated use of creative labor.
• Some insist enforcement is straightforward via audits and reproducible training; others say the real barrier is lobbying by well‑funded AI firms and rightsholders.

Human vs AI creativity

• Debates erupt over analogies between humans “trained” by life and AI trained on data.
• Many stress that humans bring lived experience, community, and emotion, while AI has none, making “it’s just like a human learning” a false equivalence.