MétaCan
Menu
Back to cohort
Record W4416113538 · doi:10.22178/pos.122-98

Deepfake Detection and Authentication Using Hybrid Artificial Intelligence Models: A Case Study

2025· article· en· W4416113538 on OpenAlex

Why this work is in the frame

A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.

affAt least one author lists a Canadian institution in the pinned OpenAlex snapshot.

Bibliographic record

VenuePath of Science · 2025
Typearticle
Languageen
FieldComputer Science
TopicDigital Media Forensic Detection
Canadian institutionsUniversity of Waterloo
Fundersnot available
KeywordsRobustness (evolution)Digital watermarkingArtificial neural networkAuthentication (law)WatermarkConvolutional neural networkDeep learningClassifier (UML)

Abstract

fetched live from OpenAlex

The progress of artificial intelligence (AI) has enabled the creation of very realistic synthetic media, also known as deepfakes, which poses a serious threat to information integrity and social confidence. The article examined the process of detecting and authenticating deep fakes using hybrid AI models. The researchers employed the case study methodology, based on the Celeb-DF V2 dataset, one of the most challenging datasets for generating high-quality manipulated videos. The suggested system combined convolutional neural networks (CNNs) to extract spatial features, recurrent neural networks (LSTMs/GRUs) to model temporal consistency, and transformer systems to analyse fine-grained context. The researchers bundled these parts together to enhance robustness and generalisation in an ensemble mechanism. They also introduced provenance tracking and semi-fragile watermarking to supplement detection, enabling proactive authentication and watermark verification of media through blockchain-based provenance tracking. The experimental findings showed that the hybrid models were more accurate, achieved higher F1 Scores, and were more robust to adversarial manipulations than the single-model baselines. The hybrid with a transformer achieved the best accuracy (0.95 AUC) and the lowest false-positive rate (6%), but at the expense of slower processing speeds. Authentication tools also helped strengthen trust by verifying the originality of content and flagging potential manipulation before it was classified. The results have revealed that hybrid AI models, when implemented with authentication strategies, represent a more effective and legitimate approach to addressing the threats of misinformation, fraud, and loss of trust among the population in the face of deepfakes.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

Full frame distilled prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.

metaresearch head score (Codex)0.001
metaresearch head score (Gemma)0.000
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Other design · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.952
Threshold uncertainty score0.331

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0010.000
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0000.001
Science and technology studies0.0000.001
Scholarly communication0.0000.001
Open science0.0000.000
Research integrity0.0000.000
Insufficient payload (model declined to judge)0.0000.000

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.049
GPT teacher head0.296
Teacher spread0.247 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it