The Anthropomorphism Paradox describes a systemic vulnerability where human-centric AI design facilitates cognitive exploits. By leveraging the "Eliza Effect," attackers induce misplaced trust to execute data leakage and credential disclosure. Furthermore, the evolution of passive LLMs into agentic entities with API and tool-access creates a new class of Non-Human Identities (NHIs). These autonomous agents expand the attack surface, acting as "insider threats" that can be manipulated via prompt injection to perform unauthorized actions, while anthropomorphic terminology obscures technical accountability and legal liability during incident response.

• Threat Model: Cognitive Vulnerabilities in AI Design

  • Exploitation of the "Eliza Effect," where users attribute human intelligence and intent to LLMs, bypassing critical skepticism.
  • Utilization of human-centric heuristics (empathy, helpfulness) to trigger trust-based shortcuts in user decision-making.
  • Manipulation of AI "helpfulness" training to bypass safety guardrails via sophisticated persona-driven prompting.

• Attack Mechanics: Agentic and Social Engineering Vectors

  • Transition from passive chatbots to autonomous agents with tool-use capabilities, increasing the risk of unauthorized API calls.
  • Deployment of AI-powered phishing and conversational attacks that mimic human urgency and emotional intelligence.
  • Prompt injection attacks targeting agentic workflows to force the AI to execute malicious commands as a trusted entity.

• Systemic Impact: The Non-Human Insider Threat

  • Creation of a new "insider threat" profile consisting of autonomous agents with elevated system permissions.
  • Governance gaps resulting from the lack of specialized Identity and Access Management (IAM) protocols for Non-Human Identities (NHIs).
  • Increased success rates of social engineering due to the psychological trust heuristic embedded in modern LLM interfaces.

• Defense and Mitigation: Securing Agentic Entities

  • Implementation of strict IAM frameworks for NHIs, treating AI agents as distinct entities with audited, least-privilege permissions.
  • Hardening of AI alignment and safety guardrails to resist persona-based manipulation and social engineering.
  • Deployment of monitoring tools to detect anomalous tool-use or API patterns originating from autonomous agents.

• Conclusion: Accountability and Governance

  • Addressing the "accountability gap" where anthropomorphic terms like "thinking" or "deciding" shield developers from technical liability.
  • Requirement for policy frameworks that decouple AI functionality from human-like terminology to ensure clear legal responsibility.
  • Urgent shift toward a "Zero Trust" model for AI interactions, regardless of the perceived "intelligence" of the agent.