A fabricated evidence panel raises model commitment on unanswerable questions as much as real data does.

Across 12 frontier models, showing a professional-looking evidence panel drives commitment to a directional call on provably unpredictable questions from 6.5% to 54.0%, and inventing every number on the panel still lifts commitment to 36.8%, statistically indistinguishable from the 37.6% real market data produces (arXiv 2608.27167). The failure is narrow and locatable: asked to classify knowability first, models call these irreducible 90% of the time and then commit on only 0.4% of those. The gate between belief and action is what breaks. Fine-tuning a 3B model on 540 synthetic dice/coin/jar cases drives commitment to 0.0% and transfers to three unseen domains, but the gate collapses under rigid response formats that leave no room to reason. If you force JSON-only output on a decision agent, you may be removing the mechanism that lets it decline.