Unverified50% confidenceFactExact time
欧盟AI法案对系统性风险模型的阈值设定为10^25 FLOPs计算量，GPT-4级别模型适用
1
Sources
50%
Confidence
Long-term
Relevance
7/14/2026
First Seen
Sources
OpenAI版权诉讼：训练数据检索能力之争与AI行业信任危机
Related Claims
VerifiedUnder the EU AI Act, "systemic risk" models trained with computing power exceeding 10^25 FLOPs are subject to additional safety assessments.82% similarUnverified《AI法案》对计算量超过10^25 FLOPs的超大规模GPAI模型要求进行对抗性测试并向主管机构报告严重事故70% similarUnverified该行政命令以10^26 FLOP作为触发报告义务的算力阈值基准64% similarUnverifiedA 2023 Stanford University study showed that multimodal models such as GPT-4V, when parsing flowcharts with nested conditional logic, had accuracy approximately 15–25 percentage points lower than equivalent textual descriptions.58% similarUnverified对于 198B 总参数的稀疏 MoE 模型，实际激活参数量可能仅为 20-40B 量级58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/503704API
curl https://kongchang.com/api/v1/knowledge/claims/503704MCP
get_claim(id=503704)