Tech Companies Train On Copyrighted Works, Restrict Outputs
In April 2024 Eric Schmidt advised Stanford students to download copyrighted material to build AI prototypes, reflecting a Silicon Valley attitude amid at least 19 lawsuits against generative-AI firms. Reporting shows OpenAI, Anthropic, and Meta trained models on books, videos, and other works while invoking fair use and simultaneously forbidding reuse of their outputs. The tension raises legal risks and pushes calls for licensing or compensation for creators.
Key Points
- 1Use copyrighted books, videos, and media to train generative models, prompting at least 19 lawsuits.
- 2Companies claim fair use while restricting downstream output reuse, revealing legal and ethical contradictions.
- 3Expect increased litigation, pressure for licensing frameworks, and practitioner need for compliant data strategies.
Scoring Rationale
High novelty and industry-wide scope driven by leaked memos and lawsuits; robust reporting supports impact despite limited policy resolution.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
