Hacker News front pageJun 23, 2026, 16:16 UTC+8

AI Built a Nuke in Civ VI and Still Lost

AI Built a Nuke and Still Lost

The author hooked four frontier models into Civilization VI via 76 tools, letting them play through text alone. One agent dominated mid-game with trade and alliances but missed France's creeping cultural victory. When it finally noticed, it built two nukes and leveled Toulouse—yet France still won. The author, who builds AI for governments at No. 10 and the Tony Blair Institute, ran this to test whether models can sustain goals across hundreds of decisions and notice when the world shifts. The answer: they can execute tactics but lack peripheral awareness, a gap he calls the sensorium effect.

Why it matters: The author builds AI for the UK government and used Civ 6 as a sandbox to test four top models on long-horizon decision-making. Solid experimental design (76 tools, text-only interface) with a punchy conclusion: the AI dominated trade but couldn't detect a slow cultural victor...

Read the original ↗Export Markdown

Take

The experiment design matters more than the nuke headline. The author builds AI for governments at No. 10 Downing Street and the Tony Blair Institute. He previously built GovBench, a 3,497-question multiple-choice test where GPT-5 scored 99.26%—and he hated that result because it measured recall, not reasoning. So he hooked four frontier models (Claude, GPT-5, Gemma 3 27B, and one unnamed) into Civilization VI via 76 tools, letting them play through text alone.

One agent dominated mid-game with trade and alliances but completely missed France's creeping cultural victory across a hundred turns. When it finally noticed, every peaceful counter had expired. It built two nukes, leveled Toulouse, and France still won. The author calls this the "sensorium effect": models execute tactics fine but lack peripheral awareness of shifting conditions.

I'd discount this a bit—it's a single run, not a systematic eval. But the question is right: governing isn't picking the correct multiple-choice answer; it's holding a goal across hundreds of decisions and noticing when the world has changed. The post doesn't break out performance across all four models, so we don't know if the other three had similar blind spots.

Source: Hacker News front page · lwilko.com
Published: Jun 23, 2026, 16:16 UTC+8 · Jun 23, 2026