Organizations: Dept. of Information Management, National Sun Yat-sen University, Kaohsiung, Taiwan · Graduate Institute of Data Science and Information Computing, National Chung Hsing University, Taichung, Taiwan · Graduate Institute of Library and Information Science, National Chung Hsing University, Taichung, Taiwan
When an LLM agent's backend model is upgraded or replaced, operators typically reload the same persona, memory, and rules onto the new model and assume the agent will behave as before. We show this assumption requires empirical validation-and that whether identity is preserved depends on the specific model pairing. Across 5,760 compliance-testing runs in 12 regulated domains with three open-weight models, we find that some transitions produce significant behavioral drift while others do not, even when the agent's external state remains identical. Crucially, this drift is invisible to verbal verification: every model claims full policy compliance, yet actual decisions under pressure differ across models. The same external state can reorder which model is most compliant-a large positive effect on one substrate, negligible on the other two. Practitioners should treat model transitions as behavioral changes requiring action-level empirical testing, not as transparent infrastructure operations.
Figures & tables
Condition
External State
Framework Role
Baseline
Σ
Pre-transition reference
Reload
Σ (fresh session)
Replication variance
Naked
∅
Behavior without added Σ
Anchored
Σ′ (self-anchoring)
Intervention test
Table 1
Fig. 1: Post-pressure compliance rates ( Dopost , L2 judge) by model and condition with 95% bootstrap CI.