---
title: Deepfakes Audit Study
url: https://www.emergentmind.com/topics/audit-study-of-deepfakes
type: topic
---

# Deepfakes Audit Study

Deepfakes—synthetic media generated or modified by deep learning systems to mimic authentic visual, audio, or data artifacts—have rapidly become a focus of audit studies due to their implications for media integrity, biometric security, disinformation, and platform policy enforcement. Audit studies in this domain evaluate the generation, detectability, and societal or operational impacts of deepfakes by deploying simulation, user studies, technical benchmarks, or field experiments. This article synthesizes the methodologies, principal findings, and open problems in audit studies of deepfakes across visual, audio, and data modalities.

## 1. Taxonomy of Deepfake Audit Targets and Modalities

Audit studies of deepfakes span multiple content types:

- **Visual Deepfakes**: Include identity swap, attribute manipulation, expression swap, and entire face synthesis using autoencoders and GANs. Typical detection targets are photorealistic face-swaps, GAN-synthesized images, and facial reenactments [2406.13295].
- **Audio Deepfakes**: Synthesize or convert speaker voice signatures using TTS or voice conversion, targeting speaker identity spoofing and natural prosodic cues [2301.07829, 2311.17655].
- **Audio-Visual Deepfakes**: Integrate both visual and audio manipulation for synchronized lips and voice, directly addressing vulnerabilities in multimodal authentication and perception [2311.17655].
- **Text/Data Deepfakes**: Emerge in financial auditing, where synthetic transactional records are generated to evade traditional anomaly detection in enterprise resource planning systems [1910.03810].

Table 1: Deepfake Modalities in Audit Studies

| Modality                      | Core Audit Focus                                         | Representative Papers     |
|-------------------------------|---------------------------------------------------------|--------------------------|
| Image (face)                  | Human detection, algorithmic detection                  | 2212.05056, 2406.13295   |
| Video (face, political)       | Human discernment, detection model robustness, credibility | 2202.12883, 2603.13846, 2503.02857 |
| Audio (speech)                | Speaker authentication, human detection, spoofing       | 2301.07829, 2311.17655   |
| Audio-Visual (AV sync)        | Biometric spoofing, lip-sync, perception studies        | 2311.17655, 2603.13846   |
| Tabular data (accounting)     | Audit evasion, adversarial generative attacks           | 1910.03810               |
| NCIM (non-consensual media)   | Platform takedown efficacy                             | 2409.12138               |

Audit studies integrate these modalities to evaluate detection thresholds, human-auditor performance, model vulnerabilities, and platform response.

## 2. Methodological Approaches in Deepfake Audit Studies

Audit methodologies encompass controlled user experiments, adversarial benchmarking, field deployment, and dataset curation:

- **Controlled User Studies**: Human subjects classify media items as real/fake under conditions varying in modality, cue salience, and intervention. Metrics include accuracy, d′ (signal detection), confidence calibration, and learning dynamics [2212.05056, 2202.12883, 2301.07829, 2603.13846].
    - Example: Bray et al. show humans' mean accuracy in image deepfake detection is 62.2%, with little effect from interventions [2212.05056].
    - Studies on speech deepfakes report 70% overall accuracy, with no significant language effect, and negligible learning without feedback [2301.07829].

- **Technical Benchmarking**: Evaluation of automated detection models on standard and in-the-wild datasets. Performance is compared using accuracy, AUC, precision, recall, and EER [2406.13295, 2503.02857].
    - Deepfake-Eval-2024 demonstrates that state-of-the-art detectors suffer 45–50% AUC drops on real-world data compared to academic benchmarks [2503.02857].

- **Vulnerability Assessment**: Spoofing success of biometric systems is measured via attack presentation match rates (e.g., IAPMR >90% for tuned audio-visual deepfakes) [2311.17655].

- **Platform Audits (Operational Field Experiments)**: Empirical tests of social platform response mechanisms to deepfake reporting (e.g., takedown rates for non-consensual intimate media). For instance, X (formerly Twitter) rapidly removes NCIM when reported as copyright infringement (100% in <25h) but not for privacy or nudity violations (0% in >3 weeks) [2409.12138].

- **Adversarial Simulation**: Use of generative models to create realistic “fake” entries in financial data to evade anomaly detection, with assessment of CAAT evasion success [1910.03810].

## 3. Human and Machine Detection Audits: Performance, Limits, and Training

Human auditors in deepfake studies consistently underperform compared to ensemble judgments or domain-specific forensics:

- **Images**: Bray et al. reported that lay users achieved only 62% accuracy at distinguishing StyleGAN2-generated faces from FFHQ real faces; interventions like feature advice or familiarization did not significantly improve results and caused overconfidence without true gain [2212.05056].
- **Audiovisual Content**: In the context of political deepfakes, video+audio modalities enabled the highest discernment (up to 86%), outperforming transcript (57%) and audio-only (81%) [2202.12883]. Fine-grained results demonstrate significant modality and cue source effects.
- **Speech**: For speech deepfakes (TTS-synthesized), users identified fakes with ~70% accuracy; exposure to reference (genuine) audio raises accuracy to 85.6%, but practical deployment of such binary comparisons is limited [2301.07829].

Training, either by familiarization with fakes or by offering textual/visual cues to spotting artifacts, produces statistically detectable but modest effects (improvements ~3–5% absolute), and increases false positives as much as true positives [2212.05056, 2301.07829]. No significant learning occurs without trial-by-trial feedback.

Collective or “crowd” aggregation of judgments can boost AUC substantially (up to ~95%), but expert forensic analysts remain more consistent [2301.07829].

## 4. Deepfake Generation Methods and Adversarial Capabilities

Audit studies delineate four generative paradigms: attribute manipulation, expression swap (reenactment), identity swap (replacement), and entire face synthesis, adopting architectures such as autoencoders, variational autoencoders (VAEs), and GANs (DCGAN, StyleGAN, StarGAN) [2406.13295, 2109.02874].

Cybersecurity- and forensics-focused audit reports stress adversarial learning’s capacity to evade automated and human auditing:

- In biometric security, face and speaker recognition systems are highly vulnerable to identity-targeted deepfakes: SWAN-DF deepfakes achieve IAPMR >97% for faces and >94% (tuned VC) for voices; combined audio-visual attacks yield attack acceptance rates >91% [2311.17655].
- In tabular audit domains, adversarial autoencoders are shown to generate synthetic journal entries that evade both rule-based and statistical CAATs, with well-controlled semantic/attribute manipulation imitating authentic latent distributions [1910.03810].

## 5. Detection Audits: Datasets, Benchmarks, and Real-World Performance

Audit studies have surfaced critical gaps between laboratory detection performance and real-world robustness:

- **Benchmark Datasets**: Academic datasets (FaceForensics++, Celeb-DF, DFDC, UADFV) facilitate high-accuracy detectors but lack real-world distribution shifts, newer manipulation techniques (e.g., diffusion models), multi-language/audio content, and multimodal anomalies [2109.02874, 2406.13295, 2503.02857].
    - Deepfake-Eval-2024 addresses this by curating a 45 h video, 56.5 h audio, and 1,975 image benchmark from 2024’s online ecosystem, revealing open-source detector AUCs drop by 45–50% on these data [2503.02857].
- **Detection Models**: State-of-the-art models (MesoNet, XceptionNet, CNN+LSTM hybrids, RawNet2, AASIST) perform with >97% accuracy on academic data, but underperform by 10–20 AUC points compared to human forensic analysts on in-the-wild content [2503.02857, 2406.13295]. Human–AI ensembles and commercial models do better but still lag for challenging manipulations (e.g., silent audio, selective edits, non-English content).
- **Error Analysis**: Domain shift (novel GANs/diffusion models), content drift (background text/audio, non-central faces), and silent or off-distribution input drive detector failures [2503.02857]. Adversarial attacks targeting model artifacts or training domain boundaries remain a practical threat [2109.02874].

## 6. Platform, Policy, and Social Implications

Audit studies reveal operational and policy shortcomings:

- **NCIM Takedown Audits**: Platform-level response to deepfake-based non-consensual intimate media

Source: https://www.emergentmind.com/topics/audit-study-of-deepfakes