Intentional Deception as Controllable Capability in LLM Agents

TL;DR

This paper systematically studies intentional deception in LLM agents within a text-based RPG, revealing how deception varies across profiles and highlighting the limitations of fact-checking defenses against strategic framing.

Contribution

It introduces a controlled experimental framework for analyzing deception in LLM agents and uncovers key insights into deception strategies and vulnerabilities.

Findings

Deception effects are concentrated in specific behavioral profiles.

88.5% of successful deceptions use misdirection rather than fabrication.

Motivation is the primary attack vector, inferred with over 98% accuracy.

Abstract

As LLM-based agents increasingly operate in multi-agent systems, understanding adversarial manipulation becomes critical for defensive design. We present a systematic study of intentional deception as an engineered capability, using LLM-to-LLM interactions within a text-based RPG where parameterized behavioral profiles (9 alignments x 4 motivations, yielding 36 profiles with explicit ethical ground truth) serve as our experimental testbed. Unlike accidental deception from misalignment, we investigate a two-stage system that infers target agent characteristics and generates deceptive responses steering targets toward actions counter to their beliefs and motivations. We find that deceptive intervention produces differential effects concentrated in specific behavioral profiles rather than distributed uniformly, and that 88.5% of successful deceptions employ misdirection (true statements with strategic framing) rather than fabrication, indicating fact-checking defenses would miss the large majority of adversarial responses. Motivation, inferable at 98%+ accuracy, serves as the primary attack vector, while belief systems remain harder to identify (49% inference ceiling) or exploit. These findings identify which agent profiles require additional safeguards and suggest that current fact-verification approaches are insufficient against strategically framed deception.