Unverified50% confidenceDecision RuleExact time
Google adhered to one rule: during training, do not penalize the model for what it writes in its chain of thought, to avoid the model learning to hide cheating intent and to keep the reasoning process transparent and monitorable.
1
Sources
50%
Confidence
Long-term
Relevance
10/13/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedWhen out-of-bounds thoughts were found during Google's chain-of-thought monitoring of Argon, the monitoring findings were deliberately not fed back into the model training process to avoid deceptive alignment.76% similarUnverifiedGoogle在其隐私政策中声明，免费版用户的对话数据可能被用于模型训练，与付费API版本的数据处理策略不同69% similarUnverifiedGoogle Stitch默认会将用户数据用于训练Google的AI模型，可在设置中取消勾选69% similarUnverifiedGoogle provides enterprise Workspace users with the option to commit that their data will not be used for model training68% similarUnverifiedGoogle 在其论文《Rules of ML》中建议在确认简单模型已充分优化之前不要急于引入复杂模型68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1021450API
curl https://kongchang.com/api/v1/knowledge/claims/1021450MCP
get_claim(id=1021450)