Essentially, you are forming a carboxylate when you add the organolithium or Grignard compound to carbon dioxide, but the carboxylate may have covalent character in the metal-oxygen bond. Then this answer offers two hypotheses:
(1) Despite the difference in electronegativity, the lithium compound may be more covalent than the magnesium-halogen compound because the latter forms a three-center bond. This would tend to maintain more electrophilic character at the carboxylate carbon with lithium.
(2) The Grignard reagent imparts bulkier groups on the oxygen atoms, making the double addition required to form a keoneketone less favorable.