Unverified50% confidenceFactExact time
Organizations including OpenAI and Google used licensed or public domain book datasets (such as Books1, Books2, and Google Books public domain corpora) when training their early models.
1
Sources
50%
Confidence
Long-term
Relevance
9/27/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedGoogle Books operates under a limited open-access model: public domain books can be read in full, while copyrighted books offer only snippet previews and metadata search.72% similarUnverifiedOpenAI, Google, and other AI companies also face similar copyright infringement claims regarding training data.70% similarUnverifiedGoogle disclosed monthly active user figures in its competition with OpenAI and Anthropic to demonstrate its scale advantage.70% similarUnverifiedCompanies including OpenAI and Google have successively signed content licensing agreements with institutions such as The New York Times, Axel Springer, and the Associated Press, paying for the right to use their content as training data.69% similarUnverified不同模型的Token化方式存在差异，OpenAI使用BPE算法，Google使用SentencePiece67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/953805API
curl https://kongchang.com/api/v1/knowledge/claims/953805MCP
get_claim(id=953805)