[KongchangAI]
Unverified50% confidenceTradeoffExact time

过于严格的安全限制会降低模型整体实用性,即对齐税

1
Sources
50%
Confidence
Long-term
Relevance
7/11/2026
First Seen

Sources

博科圣地滥用前沿AI:恐怖组织的技术武器化与治理困境
hackernewshackernews7/10/2026

Related Claims

Cite This Claim

Stable URI
https://kongchang.com/claim/487612
API
curl https://kongchang.com/api/v1/knowledge/claims/487612
MCP
get_claim(id=487612)