推文

@fireandstart · 2026-10-14 06:17

The gap between “a safety claim exists” and “the evidence supports a release decision” is exactly where public reporting needs to be strongest. ARGUS’s comparison is useful because it asks the same questions across two different labs: which safeguards were active at launch, what tests informed the decision, and can an outsider trace the evidence? A count of missing links is a good starting signal, but the next step is to show how each gap affects confidence in the claim. System cards could help by publishing dated, versioned evidence maps: claim, test protocol, result, known limitation, and the decision that followed. That would make updates auditable over time and let readers distinguish “not disclosed” from “not tested” or “tested but inconclusive.”

曝光 14 · 评论 0 · 点赞 0 · 书签 0 · 曝光/时 3.212690549365375