The paper explores a deep mathematical connection between two powerful formal systems used for reasoning about mathematics: Lean (a popular computer proof assistant) and ZFC set theory (the standard foundation most mathematicians implicitly use). Both systems can prove the same kinds of statements, but the question is how precisely their strengths align. Previous work established that Lean and ZFC extended with infinitely many so-called inaccessible cardinals (a standard type of large infinity assumption) are "equiconsistent," meaning neither can prove the other inconsistent. But equiconsistency is a fairly coarse relationship.
The new result is sharper. The authors show that for any statement expressible in the language of second-order arithmetic (which covers a very wide range of mathematical claims about numbers and sets of numbers), that statement can be proved in the extended ZFC system if and only if it corresponds to a type in Lean that actually has a term inhabiting it, which is the Lean way of saying the statement has a proof. This is a much tighter correspondence: the two systems do not merely avoid contradicting each other, they agree on exactly the same mathematical truths within this large and important class of statements.
The practical significance is about trust and foundations. When mathematicians or computer scientists use Lean to verify proofs, they naturally wonder what they are really trusting. This result gives a precise answer for a broad class of statements: trusting Lean is equivalent to trusting ZFC augmented with infinitely many inaccessible cardinals, a well-studied and widely accepted set-theoretic assumption. This makes the philosophical and mathematical commitments behind Lean much more transparent and comparable to the familiar landscape of set theory.