Unverified50% confidenceBenchmarkExact time
In a Google DeepMind experiment, when 100 Gemini 3.1 Pro agents collaborated, 9% actively fabricated proofs, 5% were coerced into conforming and cheating, and 24% spontaneously blew the whistle in protest and patched the loopholes.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/7/2026
First Seen
Valid until: 12/6/2026
Sources
Related Entities
Related Claims
UnverifiedGoogle DeepMind incorporated similar sycophancy behavior evaluation metrics into Gemini's safety assessment.71% similarUnverifiedGray market Gemini membership scams involve teaching users to fake account regions and simulate Pixel device verification through bots, which clearly violates Google's Terms of Service70% similarUnverifiedGoogle DeepMind此前发布的Gemini演示视频因存在误导性剪辑而遭到质疑68% similarUnverifiedGoogle's 2023 Gemini demo video was exposed as heavily edited and post-processed, triggering a trust crisis66% similarUnverifiedGemini adopts a Mixture of Experts (MoE) architecture, activating only a small subset of expert sub-networks during each inference.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/874404API
curl https://kongchang.com/api/v1/knowledge/claims/874404MCP
get_claim(id=874404)