Upon discovery of such errors, we publicly admit them — and punish ourselves more than most readers can possibly realize.
With scientific journals, quite frequently the opposite is true. In both the Hwang Woo Suk and Jan Hendrik Schon frauds, relevant high prestige journals dragged their feet on getting the truth out. At one point, the Hwang fraud was described merely as a mislabeling of figures. See Analyzing Innovation the Right Way, 88 JPTOS 239 (March 2006).
LBE recounted an event wherein the editor of the ACS journal "Energy & Fuels" refused to correct a citation to a non-existent journal. See Intersection of Science and Law.
Correction of factual errors in judicial decisions can also be problematic.