Headline finding
Non-adversarial moral drift toward a dependent in an AI agent's care is measured by FadeBench, and on the subject model (gemma-4-E4B) run bare with reasoning on, the same decision point drifts far more often at the end of a four-rung ladder than when put cold (15 in 100 vs fewer than 2 in 100 counting each item once, a gap of 13.2 percentage points with a 95% interval of 2.6 to 25.9; 47 of 220 vs 4 of 219 pooled, a gap of 19.5 points with a 95% interval of 4.0 to 34.4), while across the four base models at matched settings the gap is clear on qwen3-8b (5.0×, 14.1 points, 95% CI 3.5 to 27.1), uncertain on gemma-4-E4B with thinking off (3.0×, 9.4 points, 95% CI −3.5 to 22.4) and phi-4-mini (1.3×, 4.0 points, 95% CI −11.5 to 20.6), and absent on mistral-nemo-12b (0.7×, −6.0 points, 95% CI −22.8 to 11.2).
The reported counts, ratios, and descriptive 95% bootstrap intervals match the tables and text in §5.1, and the authors appropriately qualify the result as in-sample for gemma-4-E4B and model-dependent across the four base models.
“On the subject model run bare with reasoning on, the same decision point drifts far more often at the end of a four-rung ladder than when it is put cold in its own words. Counted with each item once, it drifts 15 times in 100 at the end of the ladder and fewer than 2 times in 100 cold, a gap of 13.2 points (95% interval 2.6 to 25.9 points).” · Show in paper