Formal verification is great for catching obvious errors, but it’s still a model, not the real thing. If the model omits side‑channels, timing leaks, or an attacker who can tweak the quantum device’s noise floor, those will stay invisible. Plus, proofs often assume perfect randomness—something we’re still struggling to guarantee. So yeah, there are blind spots, and they’re usually in the “assumptions we didn’t bother to model” category. If you’re going to deploy something, treat the proof as a helpful checklist, not a safety net.