InfoResearchPreprint
ChronoPatch: Black-Box Energy-Latency Attacks on Vision-Language Models
- Published
- Record updated
Summary
ChronoPatch is a black-box attack that makes vision-language models generate longer responses using only the final output. It finds visual delay anchors through response-length-guided evolutionary search, then shrinks the perturbation with subspace-calibrated decision-based optimization. Across four open-source VLMs it raised latency by 47.86%--293.34% and GPU energy use by 41.44%--249.70% under a 5% spatial budget, and it increased response length by 37.6%--81.5% on three commercial APIs.
Topics
Related items
- MediumCVE-2026-54038: LibreChat is an enhanced ChatGPT clone that supports multiple AI providers. Prior to 0.8.9, GET…Similar attack · NVD/CVE Database
- MediumCVE-2026-54045: LibreChat password reset request deletes pending email verification tokensSimilar attack · NVD/CVE Database
- MediumCVE-2026-114499: codebase-memory-mcp gitignore matcher exponential backtracking stalls indexingSimilar attack · NVD/CVE Database
- MediumCVE-2026-114497: codebase-memory-mcp provides local code intelligence for developers and AI agents through a persistent graph and MCP…Similar attack · NVD/CVE Database
- HighCVE-2026-114494: codebase-memory-mcp query_graph stack overflow via oversized Cypher RETURNSimilar attack · NVD/CVE Database