Moonshot says cyber-data filtering in Kimi K3.1 Pro’s mid-training and reinforcement learning cut its in-house exploit-test score from 90.5% to 81.0% (pass@3). Full weights are due November 4.

Published
Moonshot AI / Kimi ↗ · Source dated
—

Published
Moonshot AI / Kimi ↗ · Source dated
Over 450 Scientists Call for Default Internet Isolation of Cyber-capable AI
Google’s $10M Purchase of Spirit Airlines Data for AI Approved by Judge
OpenAI chief scientist Jakub Pachocki wants rival labs to agree on when to pause AI training.
Treasury Secretary Questions Jensen Huang's Role Amid Compliance Inquiry
China plans standards for how AI providers count tokens, bill customers and measure service quality, National Data Administration chief Liu Liehong said Nov.
Checking account…
Loading comments…