Unverified50% confidenceFactExact time
The control target of Announcement No. 48 is "frontier AI," involving multi-dimensional indicators such as model capability, training compute, and algorithm architecture, but it does not disclose specific thresholds.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/2/2026
First Seen
Valid until: 1/31/2027
Sources
Related Entities
Related Claims
UnverifiedOrdinary AI applications going overseas are not directly affected by Announcement No. 48, but exports involving top-tier model weights, training methodologies, and AI capabilities that may be used for military or surveillance scenarios will face compliance review.79% similarUnverifiedModel-layer restrictions refer to behavioral constraints built into AI providers' models at the inference or training stage, rather than relying solely on after-the-fact review at the user or platform end.76% similarUnverifiedAI will strictly implement only the features explicitly requested, and features not clearly specified may not be implemented (as demonstrated by the 404 error on the grade records module in the demo).74% similarVerifiedAn effective AI prompt requires a clear description of the goal, key constraints, style references, and sufficient domain context74% similarUnverifiedAI Agent验收的核心问题在于开发者告诉了AI'测什么'和'怎么测'，但没有明确定义'怎么才算过'的验收标准73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1125994API
curl https://kongchang.com/api/v1/knowledge/claims/1125994MCP
get_claim(id=1125994)