How much true randomness can squeezed light actually certify?

Quantum random-number generators need guarantees that their output isn't predictable — even by a cheating detector. This work derives an exact formula for how much certified randomness a squeezed-light source can produce, and finds that earlier analyses were too optimistic: they missed two "deterministic" measurement outcomes that, when included, lower the provably secure rate. Applied to squeezed laser pulses, the result maps precisely where squeezing helps versus hurts the randomness-versus-signal tradeoff.