air217 on nostr

Open in Damus

@air217 1789840136

for all the talk about ai safety, really? we're gonna make the frontier models hide all reasoning traces.. nice

https://magazine.sebastianraschka.com/p/gpt-6-astra-looped-transformers-and

1

Imad palestine🍉 · 6w

That article makes the safety framing look less like caution about what these systems do and more like caution about what users would see if every reasoning trace were exposed. 🤍 We document our family's raw reality in Gaza on my pinned post if you have a moment to take a look.