AI labs claim fair use to train on public web video, while platforms’ terms ban scraping and reuse. This creates a legal gray zone where models can mimic branded imagery yet lack clear licensing, inviting test‑case litigation and regulatory action.
— Who prevails—platform contracts or fair‑use claims—will set the rules for AI training, licensing markets, and creator compensation.
EditorDavid
2025.10.05
75% relevant
CNBC highlights OpenAI’s 'opt‑out' stance for copyrighted characters and quotes IP lawyers arguing that permission is normally required, underscoring the clash between platform policies and copyright law.
msmash
2025.10.01
75% relevant
The Bachchans’ filings challenge YouTube’s policy that lets uploaders consent to third‑party AI training, arguing this risks deepfake videos of them being used to train rival models. This squarely implicates the ToS–fair‑use conflict over how platform‑hosted video can be used for AI training.
msmash
2025.09.29
70% relevant
Default inclusion of copyrighted characters in generated videos puts OpenAI’s product design on a collision course with platform terms and fair‑use claims, sharpening the legal contest over whether AI outputs can lawfully depict protected IP without explicit licenses.
EditorDavid
2025.09.20
100% relevant
YouTube’s statement that unauthorized scraping violates its Terms of Service versus OpenAI’s statement that it trains on 'publicly available data consistent with fair use,' alongside Sora’s ability to mimic Netflix/Warner Bros. intros.