False Prophets: On the Security of World Models in Agentic Systems
cs.CR, cs.AI
Submitted: 2026-07-25
Updated: 2026-10-01
Code: https://github.com/mlsec-group/worldmodel-security
Project page: https://worldmodels.github.io
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Cosmos World Foundation Model Platform for Physical AI
- V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning
- Dream to Control: Learning Behaviors by Latent Imagination
- Targeting World Models to Compromise Robot Learning Pipelines
- Beware Untrusted Simulators -- Reward-Free Backdoor Attacks in Reinforcement Learning
- CWM: An Open-Weights LLM for Research on Code Generation with World Models
- Universal and Transferable Adversarial Attacks on Aligned Language Models
- Qwen-AgentWorld: Language World Models for General Agents
Related papers
- SoK: AI-Augmented Binary Reversing
- Relaxed Sender Anonymity for CBDC Interbank Settlement: A Zero-Knowledge Approach on Permissioned EVM
- Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages
- Efficient Fuzzy PSI under One-Sided Assumptions
- Sealing the Audit-Runtime Gap for LLM Skills
- Token Composition: A Graph Based on EVM Logs