OpenAI and Google are reported to have transcribed over a million hours of YouTube videos to train their AI models, raising concerns about potential copyright violations.
The investigation points out that OpenAI had exhausted its data sources by 2021 and resorted to transcribing podcasts, audiobooks, and YouTube videos for training new AI models.
Google's legal department was involved in broadening the scope of actions that could be taken with consumer data, indicating potential implications for privacy policies.
[
add
]
[
|
|
...
]