Generic statements (e.g., “Birds lay eggs”) express generalizations about categories. Current theories suggest that people should be especially inclined to accept generics that involve threatening information. However, previous tests of this claim have focused on generics about non‐human categories, which raises the question of whether this effect applies as readily to human categories. In Experiment 1, adults were more likely to accept generics involving a threatening (vs. a non‐threatening) property for artifacts, but this negativity bias did not also apply to human categories. Experiment 2 examined an alternative hypothesis for this result, and Experiments 3 and 4 served as conceptual replications of the first experiment. Experiment 5 found that even preschoolers apply generics differently for humans and artifacts. Finally, Experiment 6 showed that these effects reflect differences between human and non‐human categories more generally, as adults showed a negativity bias for categories of non‐human animals, but not for categories of humans. These findings suggest the presence of important, early‐emerging domain differences in people's judgments about generics.