Auditable Agents

arXiv:2604.05485 · cs.AI · Submitted 2026-04-07 · Read on arXiv

cs.AI

Submitted: 2026-04-07

Updated: 2026-09-27

Comments: 24 pages, 4 figures. Extended version. A condensed 4-page version appears in the Proceedings of the ACM AI Leadership Summit 2026 (Visionary Papers track)

Code: https://github.com/openclaw/openclaw

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Accountability — 32× in this paper · explained in 2 episodes
• Trace — 16× in this paper · explained in 2 episodes
• Benchmark — 11× in this paper · explained in 2 episodes
• Provenance — 9× in this paper · explained in 3 episodes
• Verification — 8× in this paper · explained in 2 episodes
• Alignment — 7× in this paper · explained in 1 episode
• Fidelity — 5× in this paper · explained in 2 episodes
• AI Agents — 4× in this paper · explained in 2 episodes
• Accuracy — 3× in this paper · explained in 1 episode
• Agent Architectures — 3× in this paper · explained in 1 episode
• Benchmarking — 3× in this paper · explained in 3 episodes
• Record Fidelity — 3× in this paper
• Verification Cost — 3× in this paper
• Attribution Completeness — 2× in this paper
• Audit Detection Latency — 2× in this paper
• Calibration — 2× in this paper · explained in 7 episodes
• Fine-Tuning — 2× in this paper · explained in 5 episodes
• Gap Burden — 2× in this paper
• Integrity Strength — 2× in this paper
• Action Coverage Rate — 1× in this paper
• Actor — 1× in this paper · explained in 3 episodes
• Adaptation — 1× in this paper · explained in 1 episode
• Adaptive Safety — 1× in this paper · explained in 1 episode
• Adversarial Attacks — 1× in this paper · explained in 1 episode
• Affordance — 1× in this paper · explained in 1 episode
• Agent Interactions — 1× in this paper · explained in 1 episode
• Agent Skills — 1× in this paper · explained in 1 episode
• Agentic AI — 1× in this paper · explained in 1 episode
• Aggregation — 1× in this paper · explained in 1 episode
• Attribution Chain Depth — 1× in this paper
• Core — 1× in this paper · explained in 2 episodes
• Failure Modes — 1× in this paper · explained in 3 episodes
• Formal Verification — 1× in this paper · explained in 2 episodes
• Grounding — 1× in this paper · explained in 3 episodes
• Interpretability — 1× in this paper · explained in 15 episodes
• Lifecycle Phase Coverage — 1× in this paper
• Multi-Agent System — 1× in this paper · explained in 6 episodes
• Structural Policy Decidability Rate — 1× in this paper

Sources

• AgentRx: Diagnosing AI Agent Failures from Execution Trajectories
• AgentOps: Enabling Observability of LLM Agents
• MLAgentBench: Evaluating Language Agents on Machine Learning Experimentation
• Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations
• Agent Safety Is Action Alignment
• The Autonomy Tax: Defense Training Breaks LLM Agents
• FORTIS: Benchmarking Over-Privilege in Agent Skills
• When Only the Final Text Survives: Implicit Execution Tracing for Multi-Agent Auditing
• Audit Trails for Accountability in Large Language Models
• Generative Agents: Interactive Simulacra of Human Behavior
• Gorilla: Large Language Model Connected with Massive APIs
• Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!
• ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIs
• SkillCenter: A Large-Scale Source-Grounded Skill Library for Autonomous AI Agents
• Authenticated Delegation and Authorized AI Agents
• AgentSpec: Customizable Runtime Enforcement for Safe and Reliable LLM Agents
• OpenHands: An Open Platform for AI Software Developers as Generalist Agents
• Agent Skills for Large Language Models: Architecture, Acquisition, Security, and the Path Forward
• SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering
• No Attacker Needed: Unintentional Cross-User Contamination in Shared-State LLM Agents

Related papers

• MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
• Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
• The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
• MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
• Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems
• VSAL: A Vision Solver with Adaptive Layouts for Graph Property Detection