Persistent Personas? Role-Playing, Instruction Following, and Safety in Extended Interactions

MCML Authors

Michael Hedderich

Dr.

JRG Leader Human-Centered NLP

AI and Computational Linguistics

Ali Modarressi

→ Group Hinrich Schütze
 Computational Linguistics

Hinrich Schütze

Prof. Dr.

Core PI

Computational Linguistics

Abstract

Persona-assigned large language models (LLMs) are used in domains such as education, healthcare, and sociodemographic simulation. Yet, they are typically evaluated only in short, single-round settings that do not reflect real-world usage. We introduce an evaluation protocol that combines long persona dialogues (over 100 rounds) and evaluation datasets to create dialogue-conditioned benchmarks that can robustly measure long-context effects. We then investigate the effects of dialogue length on persona fidelity, instruction-following, and safety of seven state-of-the-art open- and closed-weight LLMs. We find that persona fidelity degrades over the course of dialogues, especially in goal-oriented conversations, where models must sustain both persona fidelity and instruction following. We identify a trade-off between fidelity and instruction following, with non-persona baselines initially outperforming persona-assigned models; as dialogues progress and fidelity fades, persona responses become increasingly similar to baseline responses. Our findings highlight the fragility of persona applications in extended interactions and our work provides a protocol to systematically measure such failures.

inproceedings DHA+26

---

EACL 2026

19th Conference of the European Chapter of the Association for Computational Linguistics. Rabat, Morocco, Mar 24-29, 2026.

---

Authors

P. H. L. de Araujo • M. A. Hedderich • A. Modarressi • H. Schütze • B. Roth

---

Links

DOI

---

Research Area

 B2 | Natural Language Processing