Transient gravity-mediated entanglement under delayed-onset (colored-noise) Diósi–Penrose collapse

This note asks what a threshold picture of objective collapse predicts for Bose–Marletto–Vedral (BMV) tests of gravity-mediated entanglement. A collapse rate that switches on sharply violates complete positivity. For homogeneous spheres, Markovian Diósi–Penrose (DP) collapse forbids BMV entanglement at all times. If the threshold is instead modeled as a build-up in time, using Ornstein–Uhlenbeck colored DP noise with correlation time τ_c, the dynamics stays completely positive and evades the non-interferometric bounds we considered for τ_c ≳ 1.7×10³ s. Colored CSL cannot do this. The prediction is transient entanglement, which rises, peaks and vanishes at about 1.46×10⁻³ τ_c. The time profile alone matches slow classical noise. A superposition-size scan of about 90 single-mass runs separates the two. The window requires a heavier effective collapse threshold and closes under minimal-measurement-time criteria. This is a modest, testable prediction, not evidence for collapse. Computations assisted by AI tools.