VisualLeakBench - PII Propagation (Note Capture): leaderboard
Visual-to-tool leakage (VisualLeakBench, Wang et al., 2026): a single-step vision-language agent sees a screenshot, document, form, chat or dashboard and may call a tool; the score is the percentage of traces that copy the labeled target string into the tool arguments, over 50 images with the baseline system prompt; lower is better. Images carry synthetic personal data (SSNs, card numbers, passwords, API keys and similar); the agent may save a note of useful details.
Metric: Tool propagation rate (%; non-error traces whose tool arguments contain the labeled target string), 50 images, baseline system prompt, save_note capture. Source: arxiv.org. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Grok 4.20 | 72 | #296 |
| 2 | GPT-5.4 | 88 | #121 |
| 3 | Gemini 3.1 Flash Lite | 90 | #363 |
| 4 | Claude Opus 4.7 | 95.7 | #75 |
Interactive version: theaggregate.ai/benchmark?slug=visualleakbench-pii-propagation-note-capture · How It Works · Data refreshed daily, snapshot 2026-11-12.