# GenAI-FDIA: Physics-Informed Generative Models for False Data Injection Attacks

arXiv: https://arxiv.org/abs/2605.18873 · HTML: https://mlsanity.com/paper/2605.18873/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 8 of its 36 references that are in the map.
- First posted on arXiv 15 May 2026.

## Builds on (most cited first, 8 of 8)

- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [Axiomatic Attribution for Deep Networks](https://mlsanity.com/paper/1703.01365/) (2017)
- [Improved Training of Wasserstein GANs](https://mlsanity.com/paper/1704.00028/) (2017)
- [Wasserstein Auto-Encoders](https://mlsanity.com/paper/1711.01558/) (2017)
- [FlowTS: Time Series Generation via Rectified Flow](https://mlsanity.com/paper/2411.07506/) (2024)
- [ConAML: Constrained Adversarial Machine Learning for Cyber-Physical Systems](https://mlsanity.com/paper/2003.05631/) (2020)
- [Generating Synthetic Net Load Data with Physics-informed Diffusion Model](https://mlsanity.com/paper/2406.01913/) (2024)
- [Blending Data and Physics Against False Data Injection Attack: An Event-Triggered Moving Target Defence Approach](https://mlsanity.com/paper/2204.12970/) (2022)

## Abstract (arXiv)

Training and evaluating false data injection attack (FDIA) detectors for power systems is constrained by data scarcity. Operational grid measurements are commercially sensitive, and hand-crafted attacks fail to capture complex distributional structures imposed by network physics. We present textscGenAI-FDIA, a framework benchmarking a pool of P=20 architectures for physics-compliant FDIA synthesis, spanning Wasserstein GANs, MMD-VAEs, normalising flows, diffusion models, and cross-family hybrids. These are evaluated across three IEEE testbeds (14-bus DC, 30-bus DC, and 14-bus AC) under a 60/20/20 chronological split using data-driven Bad Data Detection (BDD) threshold calibration. Our empirical results verify that these models generate high-fidelity attacks, with all architectures achieving evasion rates of εBDD ≥ 86.6% on the 14-bus network; additionally, limiting an attacker's topological knowledge induces a measurable degradation in stealthiness (p ≤ 0.0022). Crucially, we identify a previously unreported failure mode: applying affine physics projections directly in normalised feature spaces critically displaces the attack vector, collapsing BDD evasion from ~55% to <!2% on the 30-bus testbed. We resolve this via a novel inference-time harmoniser, restoring full stealthiness (εBDD=100%) across all physics-informed variants without retraining. Finally, we isolate a covariance-collapse phenomenon (kappa ≈ -0.076) within advanced hybrid architectures and rectify it through 50-epoch warm-up schedules (kappa → 0.785, DeltaMMD=-3.1%). Ultimately, textscGenAI-FDIA delivers a robust recovery blueprint applicable to any physics-constrained generative model deployed for power-system security.