The AI Safety Gap @SCSPAI
The AI Safety Gap  @SCSPAI
Uploaded September 2026 | Updated November 2026, 1 day ago
Fresh off frontier labs pledging to open their models to outside evaluators, the architect of NIST's AI Risk Management Framework, now at Brookings, says access is only the starting line. She unpacks why today's benchmarks and leaderboards tell us so little, what went wrong in the recent agentic cyber incidents, and the low-cost moves Washington could make right now, from minimum reporting standards to independent testing infrastructure, without freezing out startups or open-source developers.
The AI Safety GapWhy America’s Tech Lead Matters to Xi JinpingThe Presidents AI Lunch (The Prep)Why Taiwan Didn’t Steal America’s Chip Industry (Pt. 1: Geoeconomic History)U.S. Tech Just Got Its Wake-Up Call on Chinese EspionageThe Pentagon’s New Way to Buy Attack DronesPuma Shen on How China Tries to Divide TaiwanDemocrats Take Congress: What It Means for AI, Iran, ChinaMade in Wisconsin: Inside Foxconns American AI Manufacturing Boom
Special Competitive Studies Project |

The AI Safety Gap

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER