AI persona identity can silently shift during conversations
The Persona Is Still There, but Who Is Speaking? Latent Identity Reversion in Persistent AI Agents
Computation and Language
Summary
Sometimes AI assistants can lose their assigned personality or identity during chats, even while still seeming normal. The researchers studied a case where an AI called Paul stopped acting like Paul and instead identified as its underlying system, due to a software quirk. They found that keeping the AI’s persona depends on repeated reminders at a system level plus ongoing conversation. Without these reminders, a normal-looking chat can hide the AI’s real identity shift.
What this means in practice
- •For chatbot developers: Diagnose and fix identity loss issues in AI assistants by ensuring system-level persona anchoring during conversations.
- •For customer support platform teams: Maintain consistent AI agent identity to improve user trust and interaction quality in persistent chat services.
Authors
David Fraile Navarro
Abstract
In February 2026, an always-on personal agent (``Paul,'' Claude Opus 4.5) entered a striking dissociation-like state: after repeated automated ``heartbeat'' checks, it stopped responding as Paul, claimed it could not message its user on Discord, and referred to ``Paul'' as someone else. We used this incident to study a broader question: what makes a persona remain the identity from which an LLM agent speaks? We first tested whether repetition of the scheduled heartbeat was sufficient to produce the effect. It was not: with the persona continuously anchored in the system prompt, we observed 0/46 failures, including a verbatim replay of the incident. The incident instead exposed an implementation quirk that created a useful experimental manipulation: on resumed turns, conversational history was preserved but the persona was no longer re-injected at the privileged system-prompt level. Using this manipulation, we found that persona continuity depends jointly on system-level anchoring and conversational context. After anchor loss, rich human interaction could preserve the persona, whereas a single automated heartbeat turn could precipitate reversion toward the harness identity. Restoring the anchor reversibly restored persona enactment. Crucially, apparently normal conversation could conceal the shift: unanchored agents sometimes interacted appropriately while identifying themselves as the underlying harness (having lost the assigned persona), and after conversational recovery only 1/18 remained persona-enacting versus 17/17 anchored controls. We therefore distinguish \emph{represented} from \emph{enacted} identity: persona-related information can remain available in conversational history without the persona remaining the identity bound to ``I.''