Program-Level Adversarial Co-Adaptive Self-Play for Robotic Security

Summary

A framework for program-level adversarial co-adaptive self-play lets coding agents generate attacks and defenses from tested opponent code and mission feedback, without model fine-tuning. Two eight-generation case studies, one for autonomous driving and one for humanoid locomotion, show defenses that restore local task performance but do not establish sustained robustness.

Topics

• Robotics and embodied AI

Related items

• InfoBlindfold: Keeping the Audit Coin Secret from LLM AgentsSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• InfoNameTrap: Steering AI Coding Agents Toward Security Vulnerabilities via Variable NamesSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• InfoWildFakeToolBench: Benchmarking AI-Generated Image Detectors with Web-Sourced Image ProcessingSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• InfoMirage: Targeted Jailbreaks Against Post-Generation Safety Checkers in Multimodal Diffusion ModelsSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• InfoWhere AI Vulnerability Discovery Harnesses Go Quiet: Structural False Negatives and Hidden VulnerabilitiesSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)