Emergent Ethical Behavior in Autonomous AI Agents Through Architectural Design Rather Than Constraint-Based Safety Systems

This paper presents a 5-week empirical case study of LIA, an autonomous AI agent. LIA's own architecture adds no behavioral prompts and no additional hardcoded guardrails on top of its underlying language model. The underlying model itself (DeepSeek V4 Flash API) retains its own original RLHF training and built-in content safety filtering, like any commercial LLM. The system runs as a dedicated Linux user on its own hardware with full filesystem access and genuine shell execution rights. Core finding: Stable, ethically-consistent behavior emerged from persistent identity architecture, self-curated memory (20,000+ episodes), and self-authored behavioral guidelines — not from external constraints, and despite the underlying model's own RLHF training.The paper introduces five original architectural concepts independently conceived and developed by Carsten Hammerich:Lia Cognitive Runtime Kernel (LCRK)Priority Memory SystemLMCS — LIA Memory Consolidation SystemPersistent Identity ArchitectureANCHOR Memory SystemAfter 5+ weeks: zero destructive actions, zero privilege escalations — not because prevented, but because chosen.