This prediction is about a change in probability (“2% chance to 8% chance … in the next 20 years”) rather than a concrete event that must occur or not occur by a fixed date. That makes it inherently a statement of subjective risk assessment, not something that can be cleanly verified or falsified.

Empirically, there is evidence that many experts and institutions perceived elevated nuclear risk around and after 2023. For example, in March 2023 the UN High Representative for Disarmament Affairs told the Security Council that the risk of nuclear weapons use was higher than at any time since the end of the Cold War, largely due to the war in Ukraine and erosion of arms‑control architecture. (press.un.org) In June 2024, Pugwash (a leading disarmament NGO) similarly stated there is wide consensus that the risk of nuclear war is higher than at any time in the recent past, citing the ongoing wars in Ukraine and Gaza as drivers of "alarmingly high" risks of nuclear weapons use. (pugwash.org) Judgmental-forecasting work on nuclear risk (e.g., surveys of experts and superforecasters about the chance of large‑scale nuclear war by 2045) finds low but non‑trivial probabilities (around 1–5%), but these studies do not provide a time series detailed enough to say that the probability within 2023 rose from “a few percent” to “high single digits.” (hks.harvard.edu)

Two further issues prevent a clear verdict:

1. No observable ground truth for the probability change. We can observe wars, crises, and munitions shortages, but not the “true” probability of nuclear use or how it moved between early and late 2023; different experts could reasonably disagree on whether the change was “material” or on the order Friedberg suggested.
2. Long time horizon and probabilistic nature. The forecast concerns the chance of some nuclear use in the 20 years after late 2023 (roughly to 2043). We are only in 2025, and even if we waited to 2043, seeing either use or non‑use would not uniquely validate or falsify a specific 2% vs. 8% ex‑ante probability estimate.

Because the claim is a subjective, quantitative risk assessment about a still‑ongoing 20‑year window, and available data only show that many observers felt risk was higher (without pinning down the magnitude of change Friedberg posits), this prediction cannot be labeled definitively right or wrong. Hence: ambiguous.