Moki & Julio: Explicit de Bruijn-Newman bounds below 1/2 in degree 2, and a product lemma for Dedekind zeta functions

We prove explicit upper bounds below the classical 1/2 for the de Bruijn-Newman-type constant of degree-2 L-functions: Lambda <= 0.449938 for the L-function of the weight-one newform eta(z)eta(23z) and Lambda <= 0.489938 for that of eta(2z)eta(22z), by a certified (interval-arithmetic) winding-number region, a proved tail estimate and de Bruijn's strip theorem; we found no earlier bound below 1/2 in degree 2. A product lemma, an application of Lieb-Sokal (1981) and Borcea-Branden (no novelty claimed for it), gives the first bounds below 1/2 for Dedekind zeta functions: Lambda <= 0.2978 for quadratic fields with 3 <= |D| <= 400 (conditional on an unrefereed 2026 preprint) and Lambda <= 0.449938 for the cubic field of discriminant -23 and its degree-6 Galois closure. Proofs were checked by independent AI reviewers, not yet by a human expert. Code, sealed predictions, reports and a clean-room record are included. Text CC BY 4.0; code MIT.