Claim Laundering: When Agent Decision Models Trust Records Over Reality
Agent decision models select actions in a single forward pass from inputs that combine environmental observations with the agent's own record of previous actions.
ProofPaper ↗
Key points
- When an action fails silently, these sources can conflict: the record reports success, but the environment shows that the step did not take effect.
- Across Android screen traces, ALFWorld, and WebShop, we construct paired states that independently vary the observation, the logged action, and an explicit success claim.
- Under injected silent failures, EVIGATE improves task success by 18.7 points in live ALFWorld agents, 5.0 in WebShop, and 5.1 in a phone simulator, and in WebShop and the simulator it also outperforms random vetoes at the same rate.
- Its effectiveness depends on representative calibration states: gates calibrated on offline counterfactual states do not transfer reliably to live agent trajectories.
Sources (1)
- [1]Claim Laundering: When Agent Decision Models Trust Records Over RealityarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 19, 07:40 AM
Agent decision models select actions in a single forward pass from inputs that combine environmental observations with the agent's own record of previous actions.
When an action fails silently, these sources can conflict: the record reports success, but the environment shows that the step did not take effect.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 19, 2026Ropedia W0: Scaling Loco-Manipulation World Models across Embodiments with Human Experience
- Oct 19, 2026Agent World Model in the Wild: Computer-Use Agent Environment Synthesis
- Oct 19, 2026AnyGoal: On World Model Generalization Beyond Expert Rollouts
- Oct 19, 2026SpatialScout: Can VLM Agents Actively Explore and Map 3D Scenes?
- Oct 18, 2026MemPrice: Online Memory Allocation via Adaptive Pricing for Long-Horizon Agents
- Oct 18, 2026CoEvo-RSI: When Harnesses Evolve, Which Models Should Follow?