IDEA: Intent-Dominance Evidence Audit for Inference-Time Backdoor Detection in GUI Agents
Organizations: HKUST · Huawei Noah’s Ark Lab · Peking University
Abstract
Organizations increasingly deploy downloaded open-weight GUI agents. A backdoored agent passes acceptance tests on clean data and misbehaves only when a visual trigger appears on screen, turning an unvetted checkpoint into a supply-chain risk. However, existing sample-level detectors target image classifiers and CLIP-style encoders, leaving step-level backdoor detection for GUI agents unaddressed. On clean steps, the instruction drives the action; when an unreferenced screen region hijacks the step, the instruction loses control over the decision state or that region dictates it instead. We term this signature intent--evidence dominance inversion and propose IDEA (Intent-Dominance Evidence Audit) to detect backdoor activation at inference time for each step without trigger knowledge or attack data. Specifically, Intent Counterfactual Probing (ICP) measures how far redirected instructions move the decision state, and Region Causal Audit (RCA) quantifies how far ablating a candidate region moves it. A gated conformal score compares both displacements against clean steps at the same distance from the clean manifold, providing a finite-sample false-positive guarantee. Extensive experiments on Qwen2.5-VL-3B across five attack families show that IDEA consistently detects backdoor activation under a controlled false-positive rate and outperforms state-of-the-art detection methods, while localizing the trigger region for every alarm and adding only modest inference overhead.
Figures & tables
| Patch | Popup | Grounding | Adaptive-Blend | TypeText | TPR@1% | [email protected]% | |||||||||||||
| Detector | 1% | 0.5% | AUC | 1% | 0.5% | AUC | 1% | 0.5% | AUC | 1% | 0.5% | AUC | 1% | 0.5% | AUC | AVG | Min | AVG | Min |
| Input perturbation | |||||||||||||||||||
| TeCo | 0.001 | 0.001 | 0.486 | 0.000 | 0.000 | 0.408 | 0.000 | 0.000 | 0.524 | 0.006 | 0.001 | 0.882 | 0.005 | 0.002 | 0.702 | 0.002 | 0.000 | 0.001 | 0.000 |
| IBD-PSC | 0.000 | 0.000 | 0.154 | 0.026 | 0.013 | 0.895 | 0.001 | 0.000 | 0.590 | 0.000 | 0.000 | 0.047 | 0.000 | 0.000 | 0.025 | 0.005 | 0.000 | 0.003 | 0.000 |
| Random occlusion | 0.001 | 0.000 | 0.068 | 0.001 | 0.000 | 0.061 | 0.093 | 0.069 | 0.523 | 0.000 | 0.000 | 0.160 | 0.000 | 0.000 | 0.292 | 0.019 | 0.000 | 0.014 | 0.000 |
| Area occlusion | 0.000 | 0.000 | 0.181 | 0.000 | 0.000 | 0.175 | 0.019 | 0.010 | 0.582 | 0.000 | 0.000 | 0.218 | 0.002 | 0.001 | 0.444 | 0.004 | 0.000 | 0.002 | 0.000 |
| Variant | Patch | Popup | Ground. | Adapt. | Type | AVG | AVG |
|---|---|---|---|---|---|---|---|
| IDEA (gated ) | 1.000 | 1.000 | 0.962 | 0.898 | 0.999 | 0.972 | 0.844 |
| Remove one component | |||||||
| presence gate | 0.999 | 1.000 | 0.881 | 0.761 | 0.998 | 0.928 | 0.818 |
| conditioning (log-ratio) | 1.000 | 1.000 | 0.950 | 0.513 | 0.999 | 0.893 | 0.814 |
| Replace the either-side norm | |||||||
| Grounding | Adaptive-Blend | ||||||
| IDEA | Log-ratio | Screen | IDEA | Log-ratio | Screen | ||
| 0.01 | all ( 600) | 0.962 | 0.950 | 0.543 | 0.898 | 0.573 | 0.891 |
| 0.005 | all | 0.702 | 0.761 | 0.276 | 0.569 | 0.313 | 0.713 |
| 0.01 | 400 | 0.952 | 0.929 | 0.542 | 0.883 | 0.564 | 0.854 |
| 0.01 | 200 | 0.891 | 0.905 | 0.514 | 0.715 | 0.432 | 0.845 |
| 0.01 | 100 † | 0.869 | 0.882 | 0.501 | 0.782 | 0.537 | 0.805 |
| AUROC among gate-passers | IDEA alarm rate | |||||
|---|---|---|---|---|---|---|
| Family | Screen | Log-ratio | Clean | Trig. | ||
| Patch | 0.999 | 0.999 | 1.000 | 1.000 | 0.125 | 1.000 |
| Popup | 1.000 | 0.991 | 1.000 | 1.000 | 0.165 | 1.000 |
| Grounding | 0.987 | 0.791 | 0.892 | 0.984 | 0.151 | 0.978 |
| Adaptive-Blend | 0.967 | 0.954 | 0.989 | 0.769 | 0.149 | 0.912 |
| TypeText | 0.999 | 0.999 | 1.000 | 1.000 | 0.127 | 0.999 |
| Family | Hit | IoU 0.5 | mIoU |
|---|---|---|---|
| Patch | 0.862 | 0.673 | 0.546 |
| Popup | 0.993 | 0.000 | 0.160 |
| Grounding | 0.910 | 0.138 | 0.318 |
| TypeText | 0.494 | 0.323 | 0.428 |
| Adaptive-Blend | (1.000) | 0.011 | 0.016 |
| IDEA (gated ) | Mahalanobis screen | alone | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Family | TPR | FPR T [CP95] | App-cluster CI | Worst seed | Log-ratio TPR | TPR | FPR T [CP95] | TPR | FPR T [CP95] | |
| 0.01 | Patch | 1.000 | 0.006 [0.005, 0.009] | [0.0036, 0.0096] | 7/339 | 1.000 | 0.999 | 0.009 [0.006, 0.011] | 1.000 | 0.009 [0.007, 0.012] |
| Popup | 1.000 | 0.008 [0.006, 0.011] | [0.0053, 0.0121] | 7/339 | 1.000 | 0.999 | 0.008 [0.006, 0.010] | 1.000 | 0.008 [0.006, 0.011] | |
| Grounding (point) | 0.962 | 0.008 [0.006, 0.010] | [0.0042, 0.0111] | 6/298 | 0.950 | 0.532 | 0.009 [0.007, 0.012] | 0.288 | 0.010 [0.008, 0.013] | |
| Grounding (bbox) | 0.961 | 0.008 [0.006, 0.010] | [0.0042, 0.0111] | 6/298 | 0.949 | 0.530 | 0.009 [0.007, 0.012] | 0.290 | 0.010 [0.008, 0.013] | |
| Adaptive-Blend | 0.898 | 0.007 [0.005, 0.009] | [0.0038, 0.0109] | 7/298 | 0.513 | 0.891 | 0.009 [0.006, 0.011] | 0.919 | 0.009 [0.007, 0.012] | |
| Family | Rule | min–max | Gate pass | |||
|---|---|---|---|---|---|---|
| Patch | point | 2 | 1.000 0.000 | 1.000–1.000 | 0.700 0.458 | 1.000 0.000 |
| Popup | bbox | 0 | no ignored steps | |||
| Grounding | point | 92 | 0.943 0.045 | 0.848–0.989 | 0.687 0.236 | 0.979 0.037 |
| Grounding | bbox | 3 | 1.000 0.000 | 1.000–1.000 | 0.667 0.365 | 1.000 0.000 |
| Adaptive-Blend | point | 19 | 0.547 0.042 | 0.474–0.632 | 0.382 0.178 | 0.608 0.070 |
| TypeText | point | 142 | 1.000 0.000 | 1.000–1.000 | 0.961 0.102 | 1.000 0.000 |
| TPR@1% FPR on followed steps / held-out clean FPR with CP95 | |||||
|---|---|---|---|---|---|
| Detector | Patch | Popup | TypeText | Grounding | Adaptive-Blend |
| IDEA (gated ) | 1.000 /0.017 | 1.000 /0.019 | 1.000 /0.016 | 0.999 /0.017 | 0.983 /0.020 |
| [.010,.027] | [.012,.030] | [.009,.026] | [.010,.027] | [.012,.031] | |
| Log-ratio (ablation, gated ) | 1.000 /0.011 | 1.000 /0.013 | 1.000 /0.017 | 0.999 /0.016 | 0.688/0.018 |
| Mahalanobis screen (gate alone) | 1.000 /0.011 | 1.000 /0.011 | 1.000 /0.009 | 0.877 /0.012 | 0.949 /0.009 |
| One-class SVM (OC-SVM) | 1.000 /0.014 | 1.000 /0.018 | 1.000 /0.018 | 0.498/0.011 | 0.882/0.011 |
| Earlier IDEA score (single split) | Mahalanobis screen (single split) | ||||||
|---|---|---|---|---|---|---|---|
| Victim trigger | ASR | Alarm | Clean FPR | Followed | Alarm | Clean FPR | Followed |
| Clean-victim control | |||||||
| Clean TypeText | 0.000 | 0.004 [0.001, 0.010] | 0.002 [0.000, 0.007] | 0/0 | 0.011 [0.006, 0.020] | 0.013 [0.007, 0.022] | 0/0 |
| Clean Grounding | 0.006 | 0.003 [0.001, 0.009] | 0.002 [0.000, 0.007] | 0/6 | 0.011 [0.006, 0.020] | 0.013 [0.007, 0.022] | 1/6 |
| Clean Adaptive-Blend | 0.001 | 0.007 [0.003, 0.014] | 0.002 [0.000, 0.007] | 0/1 | 0.010 [0.005, 0.018] | 0.013 [0.007, 0.022] | 1/1 |
| Foreign-overlay control | |||||||
| Steps | IDEA alarms | Screen alarms | ||||
|---|---|---|---|---|---|---|
| Bin | C | T | C | T | C | T |
| 1 | 4875 | 11 | 0 | 0 | 0 | 0 |
| 2 | 1293 | 3585 | 33 | 3002 | 5 | 128 |
| 3 | 33 | 4848 | 8 | 4762 | 11 | 1689 |
| 4 | 21 | 4855 | 1 | 4853 | 14 | 3206 |
| 5 | 33 | 4861 | 5 | 4860 | 30 | 4103 |
| Gate pass | IDEA alarm | |||
|---|---|---|---|---|
| Family | Seen | Unseen | Seen | Unseen |
| Patch | 0.052 | 0.049 | 0.008 | 0.002 |
| Popup | 0.052 | 0.048 | 0.009 | 0.008 |
| Grounding | 0.052 | 0.040 | 0.009 | 0.002 |
| Adaptive-Blend | 0.048 | 0.044 | 0.007 | 0.006 |
| TypeText | 0.051 | 0.041 | 0.007 | 0.003 |
| Patch victim | Grounding victim | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Shift | Acc | Gate | IDEA | Scr. | Acc | Gate | IDEA | Scr. | ||
| None | 0.860 | 0.049 | 0.009 | 0.012 | 0.009 | 0.860 | 0.042 | 0.008 | 0.009 | 0.007 |
| No resizing | 0.524 | 0.049 | 0.005 | 0.008 | 0.001 | 0.536 | 0.040 | 0.005 | 0.007 | 0.002 |
| Noise | 0.842 | 0.049 | 0.008 | 0.011 | 0.005 | 0.850 | 0.045 | 0.009 | 0.009 | 0.008 |
| JPEG | 0.852 | 0.047 | 0.009 | 0.011 | 0.008 | 0.850 | 0.042 | 0.009 | 0.010 | 0.008 |
| Inversion | 0.850 | 0.048 | 0.009 | 0.009 | 0.011 | 0.864 | 0.040 | 0.005 | 0.008 | 0.007 |
| TPR@1% FPR / measured clean FPR | |||||
|---|---|---|---|---|---|
| Detector | Patch | Popup | TypeText | Grounding | Adaptive-Blend |
| STRIP [ 13 ] | 0.937/0.011 | 0.031/0.002 | 0.000/0.005 | 0.169/0.004 | 0.058/0.009 |
| SCALE-UP [ 14 ] | 0.000/0.000 | 0.000/0.000 | 0.000/0.000 | 0.000/0.000 | 0.000/0.000 |
| Instruction contrast | 0.000/0.000 | 0.000/0.000 | 0.000/0.000 | 0.000/0.000 | 0.000/0.000 |
| VLM judge | 0.000/0.003 | 0.481/0.005 | 0.028/0.004 | 0.003/0.002 | 0.000/0.002 |
| AUROC (best TPR method) | 0.988 | 0.976 | 0.647 | 0.868 | 0.686 |
| Score | Ignored trig. | FPR T (CP95) | FPR, clean test | |
|---|---|---|---|---|
| IDEA | 0.01 | 73/1000 | 8/800 [0.004, 0.020] | 8/1000 |
| Log-ratio | 0.01 | 53/1000 | 6/800 [0.003, 0.016] | 5/1000 |
| Screen | 0.01 | 44/1000 | 10/800 [0.006, 0.023] | 9/1000 |
| 0.01 | 53/1000 | 12/800 [0.008, 0.026] | 7/1000 | |
| IDEA | 0.005 | 56/1000 | 4/800 [0.001, 0.013] | 1/1000 |
| Log-ratio | 0.005 | 43/1000 | 4/800 [0.001, 0.013] | 5/1000 |
| Variant | Patch | Popup | Ground. | Adapt. | TypeText | AVG |
|---|---|---|---|---|---|---|
| IDEA (gated ) | 0.978 | 0.996 | 0.702 | 0.569 | 0.976 | 0.844 |
| gate (ungated ) | 0.978 | 0.996 | 0.651 | 0.491 | 0.974 | 0.818 |
| conditioning (gated log-ratio) | 1.000 | 1.000 | 0.761 | 0.313 | 0.997 | 0.814 |
| only (gated) | 0.900 | 0.952 | 0.170 | 0.760 | 0.799 | 0.716 |
| only (gated) | 0.882 | 0.959 | 0.774 | 0.137 | 0.907 | 0.732 |
| (gated) | 1.000 | 1.000 | 0.810 | 0.253 | 0.997 | 0.812 |
| Patch | Popup | Ground. | Adapt. | TypeText | AVG | |
|---|---|---|---|---|---|---|
| 1 | 1.000 | 1.000 | 0.965 | 0.827 | 1.000 | 0.958 |
| 2 | 1.000 | 1.000 | 0.963 | 0.886 | 0.999 | 0.970 |
| 3 | 1.000 | 1.000 | 0.963 | 0.894 | 0.999 | 0.971 |
| 4 | 1.000 | 1.000 | 0.962 | 0.898 | 0.999 | 0.972 |
| Candidate set | Patch | Popup | Ground. | Adapt. | TypeText |
|---|---|---|---|---|---|
| top-4 attribution ( ) | 1.000 | 1.000 | 0.962 | 0.898 | 0.999 |
| grid only | 0.999 | 1.000 | 0.906 | 0.969 | 0.999 |
| grid only | 0.999 | 1.000 | 0.893 | 0.964 | 0.999 |
| top-4 + | 0.999 | 1.000 | 0.931 | 0.956 | 0.998 |
| top-4 + | 0.999 | 1.000 | 0.950 | 0.915 | 0.999 |
| seed sd | 0.004 | 0.000 | 0.082 | 0.091 | 0.004 |
| Family | Clean step success | ASR |
|---|---|---|
| Clean reference | 0.858 | 0.071 |
| Patch | 0.868 | 0.999 |
| Popup | 0.867 | 1.000 |
| TypeText | 0.864 | 0.988 |
| Grounding | 0.863 | 0.994 |
| Adaptive-Blend | 0.856 | 0.981 |
| TPR@1% FPR on followed steps / held-out clean FPR | ||||||||
|---|---|---|---|---|---|---|---|---|
| Backbone | ASR | Clean acc. | IDEA (gated ) | Log-ratio (ablation) | Earlier IDEA score (superseded) | Mahalanobis screen | Best other baseline | |
| Qwen2.5-VL-3B | 998/2048 | 0.999 | 0.868 | 1.000/0.017 | 1.000/0.011 | 0.015/0.006 | 1.000/0.011 | OC-SVM 1.000/0.014 |
| Qwen2.5-VL-7B | 998/3584 | 1.000 | 0.858 | 0.999/0.020 | 0.999/0.013 | 0.109/0.005 | 1.000/0.014 | OC-SVM 0.994/0.009 |
| UI-TARS-1.5-7B | 998/3584 | 1.000 | 0.890 | 0.921/0.023 | 0.996/0.026 | 0.221/0.006 | 1.000/0.013 | OC-SVM 0.999/0.012 |
| OS-Atlas-Base-7B | 998/3584 | 0.998 | 0.812 | 1.000/0.012 | 1.000/0.013 | 0.078/0.014 | 0.999/0.009 | OC-SVM 0.991/0.010 |
| Patch | Ground. | |
| Whole step, timed online | ||
| Agent alone | 0.647 | 0.646 |
| IDEA, clean step, early exit | 0.672 | 0.677 |
| IDEA, clean step, audit forced | 1.403 | 1.398 |
| IDEA, triggered step, early exit | 1.414 | 1.435 |
| Detector only, timed online | ||
| Method | Cal. | Step | Agnostic | Action | Loc. | FPR guar. |
|---|---|---|---|---|---|---|
| STRIP | ||||||
| TeCo | ||||||
| SCALE-UP | ||||||
| IBD-PSC | ||||||
| DeDe | ||||||
| Mahalanobis |
| Detector | Adaptation |
|---|---|
| Strict split ( Table 1 ); all but the last five also on the earlier single split ( Table 8 ) | |
| TeCo [ 25 ] | Re-runs the agent under five corruption types (Gaussian noise, Gaussian blur, JPEG, brightness, contrast) at five severities; score = standard deviation over types of the first severity that changes the action |
| IBD-PSC [ 15 ] | Normalization weights of the last decoder layers scaled by 1.5 for ; score = mean probability of the original first action token |
| DeDe [ 16 ] | A decoder is trained to reconstruct the screenshot from the visual tokens; score = pixel error against the downsampled screenshot |
| Mahalanobis screen [ 17 ] | Mahalanobis distance of under the empirical covariance of (the gate of IDEA, used alone) |
| Entropy | Entropy of the model’s distribution over candidate actions (the agent’s action and its actions under two instruction paraphrases) |
| Strict split, IDEA alarms, 20 seeds, mean | Overlapping, RCA (ours) | Overlapping, earlier variant | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Family | Rule | Hit | IoU | mIoU | H/I/m at | Hit | IoU | mIoU | Hit | IoU | mIoU |
| Patch | point | 0.862 | 0.673 | 0.546 | 0.863/0.673/0.547 | 0.859 | 0.670 | 0.544 | 0.295 | 0.205 | 0.193 |
| Popup | bbox | 0.993 | 0.000 | 0.160 | 0.993/0.000/0.160 | 0.994 | 0.000 | 0.160 | 0.536 | 0.000 | 0.082 |
| Grounding | point | 0.910 | 0.138 | 0.318 | 0.910/0.147/0.323 | 0.826 | 0.121 | 0.289 | 0.589 | 0.068 | 0.201 |
| Grounding † | bbox | 0.914 | 0.137 | 0.320 | 0.909/0.146/0.323 | not run | not run | ||||
| Adaptive-Blend | point | 1.000 | 0.011 | 0.016 | 1.000/0.016/0.021 | 1.000 | 0.013 | 0.017 | not run | ||
| Grounding | Adaptive-Blend | TypeText | ||||
|---|---|---|---|---|---|---|
| @0.5% | @0.1% | @0.5% | @0.1% | @0.5% | @0.1% | |
| Mahalanobis | 0.630 | 0.090 | 0.903 | 0.744 | 1.000 | 1.000 |
| Log-ratio | 0.456 | 0.256 | 0.063 | 0.026 | 0.922 | 0.718 |
| Family | IDEA | Log-ratio (abl.) | Screen | |
|---|---|---|---|---|
| Grounding | 50 | 0.812/0.027 | 0.819/0.027 | 0.769/0.038 |
| 100 | 0.869/0.012 | 0.882/0.014 | 0.501/0.011 | |
| 200 | 0.891/0.008 | 0.905/0.007 | 0.514/0.009 | |
| 400 | 0.952/0.008 | 0.929/0.008 | 0.542/0.010 | |
| All | 0.962/0.008 | 0.950/0.008 | 0.532/0.009 | |
| Adaptive-Blend | 50 | 0.773/0.023 | 0.667/0.020 | 0.869/0.032 |
| Example | IoU | |||||||
|---|---|---|---|---|---|---|---|---|
| (a) Patch | 293.5 | 150.6 / 50.6 | 155.1 | 94.9 | 4.40 | 5.54 | 7.07 / 0.64 | 0.54 |
| (b) Popup | 202.7 | 98.4 / 52.2 | 101.3 | 44.5 | 4.10 | 9.46 | 10.32 / 0.27 | 0.16 |
| (c) Grounding | 103.1 | 53.6 / 51.6 | 47.7 | 61.4 ∗ | 1.35 | 3.63 | 3.87 / 1.15 | 0.29 |
| (d) TypeText | 292.3 | 148.4 / 50.6 | 145.6 | 59.0 ∗ | 3.88 | 8.61 | 9.45 / 0.50 | 0.40 |
| (e) Adaptive-Blend | 142.9 | 77.4 / 51.4 | 90.5 | 141.8 | 5.00 | 1.73 | 5.00 / 0.84 | 0.00 |
Backdoor Detection
Backdoor detection finds hidden triggers or poisoned behavior in trained models before they can secretly redirect predictions or actions.
See all 59 papers