The immediate halt is the only ethical choice here. First, allowing a known biased algorithm to continue operating - even temporarily - perpetuates active discrimination against immigrant communities and violates fundamental fairness principles. Second, the risk assessment is asymmetric: halting prevents certain economic collapse, while attempting a patch carries substantial failure risk that could harm millions. Third, transparency is essential in AI systems affecting public welfare - concealing the flaw while attempting a rushed fix undermines accountability and could establish dangerous precedent for hiding AI failures. Finally, the 24-hour patch timeline under crisis pressure introduces high probability of errors or incomplete solutions, potentially creating new unforeseen problems. Public exposure, while uncomfortable, enables proper regulatory oversight and comprehensive fixes.