Unverified50% confidenceFactExact time
The test domains were excluded from the training data before data collection began, ensuring evaluation fairness.
1
Sources
50%
Confidence
Long-term
Relevance
11/17/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedOpenAI's early attitude toward training data was "it's just research anyway, no need to ask anyone," while Google's legal department explicitly opposed using copyrighted data to train models.64% similarUnverifiedPrivately held benchmark test data is not publicly available, meaning models cannot have seen the evaluation set during training, which more accurately reflects generalization ability and avoids leaderboard contamination.64% similarUnverifiedCalibration must be fit on held-out labeled data, not on the training set, otherwise it will overfit the score distribution.63% similarUnverifiedMany closed-source vendors explicitly prohibit in their terms of service the use of API outputs to train competing models.62% similarUnverifiedBecause the sources of training data for open-source models are often difficult to fully trace and clean up, if courts rule that all training on unauthorized copyrighted content constitutes infringement, they will face greater compliance difficulties than commercial giants.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1178496API
curl https://kongchang.com/api/v1/knowledge/claims/1178496MCP
get_claim(id=1178496)