CuraSec

tag: Vision-Language-Models · 2 items

Learn Meta-Adaptive Jailbreaking Achieves 80%+ ASR Against Frontier VLMs
2026-08-31 · arXiv cs.CR · source ↗ #llm-security#jailbreaking#vision-language-models
  • Engineer — Learn: If your product integrates GPT-4o, Gemini, or similar multimodal models, this research shows existing content-safety wrappers are brittle against adaptive attackers; no patch exists yet, but it motivates evaluating your VLM endpoints against adaptive prompt-injection test suites.
  • SOC/IR — Skip
  • Leader — Learn: Research demonstrating high-success jailbreaks against GPT-4o and Gemini is useful framing for board-level AI risk discussions and for questioning AI vendor safety assurance claims when procuring or expanding VLM-based tooling.
Learn archived Adversarial Attacks Can Silently Manipulate VLM Confidence Scores
2026-08-10 · arXiv cs.CR · source ↗ #ai-security#adversarial-ml#vision-language-models
  • Engineer — Learn: White-box attacks can degrade confidence readouts in vision-language models to near-random while leaving the generated answer unchanged, undermining confidence-gated pipelines; teams deploying VLMs with confidence thresholds for access control or oversight should treat confidence as an untrusted signal in adversarial contexts.
  • SOC/IR — Skip
  • Leader — Learn: Academic research showing that AI confidence gating — a common oversight mechanism in deployed vision-language products — can be silently subverted; worth tracking as AI governance frameworks and internal AI-use policies mature, but no immediate action warranted.