Unverified50% confidenceFactExact time
HC at major companies is basically frozen, except that DeepSeek and Qwen are hiring somewhat more. But the freeze is not uniform: foundation model teams take on any good talent that comes along, and what is really contracting is the application side.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/19/2026
First Seen
Valid until: 11/2/2026
Sources
Related Entities
Related Claims
UnverifiedMoE models activate only a few experts per inference while most experts sit idle, and this characteristic of "large total, small activation" is naturally suited to placing cold data on slow but high-capacity storage media.63% similarUnverified冻结基准可能因数据泄露被纳入模型训练数据而随时间失去区分力，这是一个权衡63% similarUnverified单方面放慢AI研发可能导致能力真空被约束更少的参与者填补，涉及经典的安全困境博弈论问题62% similarUnverifiedClosed-loop success rates systematically overestimate the rankability of frozen latent representations.62% similarUnverifiedDomestic or open-source models such as GLM and DeepSeek have fewer restrictions on tasks like vulnerability analysis, partly due to differences in training data and alignment strategies, but may be more permissive in other safety boundaries.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1067014API
curl https://kongchang.com/api/v1/knowledge/claims/1067014MCP
get_claim(id=1067014)