Unverified50% confidenceFactExact time
The Abliteration technique was first popularized by community researchers including failspy in 2024.
1
Sources
50%
Confidence
Long-term
Relevance
9/2/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAnthropic在2023—2024年在可解释性研究中通过稀疏自编码器取得里程碑式进展62% similarUnverified谄媚行为问题最早在2022年前后被AI对齐研究社区系统性地提出61% similarUnverified智能体概念的学术根源可追溯至1990年代分布式人工智能研究，2023年 AutoGPT、BabyAGI 等项目引发社区热潮后真正爆发60% similarUnverifiedAnthropic在2024年发布的技术路线图中将Agentic能力列为核心发展方向，并推出了支持工具调用和计算机操作的API接口59% similarUnverifiedSince 2024, the community has seen multiple incidents caused by AI tools automatically executing rm -rf or incorrectly modifying configuration files.58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/845778API
curl https://kongchang.com/api/v1/knowledge/claims/845778MCP
get_claim(id=845778)