# Learning as Deepfakes Evolve: RF-Prompt for Continual Audio Deepfake Detection

arXiv: https://arxiv.org/abs/2609.37586 · HTML: https://mlsanity.com/paper/2609.37586/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 27 of its 32 references that are in the map.
- First posted on arXiv 29 Sep 2026.

## Builds on (most cited first, 27 of 27)

- [Overcoming catastrophic forgetting in neural networks](https://mlsanity.com/paper/1612.00796/) (2016)
- [Learning to Prompt for Continual Learning](https://mlsanity.com/paper/2112.08654/) (2021)
- [DualPrompt: Complementary Prompting for Rehearsal-free Continual Learning](https://mlsanity.com/paper/2204.04799/) (2022)
- [CODA-Prompt: COntinual Decomposed Attention-based Prompting for Rehearsal-Free Continual Learning](https://mlsanity.com/paper/2211.13218/) (2022)
- [Seed-TTS: A Family of High-Quality Versatile Speech Generation Models](https://mlsanity.com/paper/2406.02430/) (2024)
- [Continual Learning of Context-dependent Processing in Neural Networks](https://mlsanity.com/paper/1810.01256/) (2018)
- [ASVspoof 2019: Future Horizons in Spoofed and Fake Audio Detection](https://mlsanity.com/paper/1904.05441/) (2019)
- [Qwen3-TTS Technical Report](https://mlsanity.com/paper/2601.15621/) (2026)
- [ADD 2022: the First Audio Deep Synthesis Detection Challenge](https://mlsanity.com/paper/2202.08433/) (2022)
- [One-class Learning Towards Synthetic Voice Spoofing Detection](https://mlsanity.com/paper/2010.13995/) (2020)
- [The Codecfake Dataset and Countermeasures for the Universally Detection of Deepfake Audio](https://mlsanity.com/paper/2405.04880/) (2024)
- [Continual Learning for Fake Audio Detection](https://mlsanity.com/paper/2104.07286/) (2021)
- [ASVspoof 5: Design, Collection and Validation of Resources for Spoofing, Deepfake, and Adversarial Attack Detection Using Crowdsourced Speech](https://mlsanity.com/paper/2502.08857/) (2025)
- [Attentive Merging of Hidden Embeddings from Pre-trained Speech Model for Anti-spoofing Detection](https://mlsanity.com/paper/2406.10283/) (2024)
- [What to Remember: Self-Adaptive Continual Learning for Audio Deepfake Detection](https://mlsanity.com/paper/2312.09651/) (2023)
- [Componential Prompt-Knowledge Alignment for Domain Incremental Learning](https://mlsanity.com/paper/2505.04575/) (2025)
- [Do You Remember? Overcoming Catastrophic Forgetting for Fake Audio Detection](https://mlsanity.com/paper/2308.03300/) (2023)
- [RainbowPrompt: Diversity-Enhanced Prompt-Evolving for Continual Learning](https://mlsanity.com/paper/2507.22553/) (2025)
- [Generalizable Audio Deepfake Detection via Latent Space Refinement and Augmentation](https://mlsanity.com/paper/2501.14240/) (2025)
- [Prompt Tuning for Audio Deepfake Detection: Computationally Efficient Test-time Domain Adaptation with Limited Target Dataset](https://mlsanity.com/paper/2410.09869/) (2024)
- [One-Class Learning with Adaptive Centroid Shift for Audio Deepfake Detection](https://mlsanity.com/paper/2406.16716/) (2024)
- [Region-Based Optimization in Continual Learning for Audio Deepfake Detection](https://mlsanity.com/paper/2412.11551/) (2024)
- [Towards Robust Audio Deepfake Detection: A Evolving Benchmark for Continual Learning](https://mlsanity.com/paper/2405.08596/) (2024)
- [One-Prompt Strikes Back: Sparse Mixture of Experts for Prompt-based Continual Learning](https://mlsanity.com/paper/2509.24483/) (2025)
- [Is Prompt Selection Necessary for Task-Free Online Continual Learning?](https://mlsanity.com/paper/2604.04420/) (2026)
- [AT-ADD: A Benchmark and Challenge for Robust and All-Type Audio Deepfake Detection](https://mlsanity.com/paper/2608.23437/) (2026)
- [GAP-Prompt: Gated Adaptive Prompting for Efficient Continual Learning](https://mlsanity.com/paper/2608.23782/) (2026)

## Abstract (arXiv)

Continual audio deepfake detection requires learning newly emerging deepfake methods while retaining discrimination of previously encountered speech. Existing dataset-incremental evaluation changes both real-speech domains and deepfake mechanisms, making their effects difficult to distinguish. We construct five task organizations over identical training, development, and evaluation pools to study these factors under a controlled sample budget. Our proposed Real-Anchored Mechanism-Incremental (RAMI) protocol reflects the practical setting in which available real speech provides a recurring mixed-domain reference while new deepfake mechanisms arrive incrementally. We further propose RF-Prompt, an asymmetric continual prompt-learning method that preserves reusable real-speech knowledge through a shared real prompt and expands mechanism-specific knowledge through inherited fake experts with orthogonal residuals. Input-adaptive soft fusion combines the accumulated experts into a fixed number of injected tokens without requiring task identity at inference. On RAMI, RF-Prompt achieves 10.110% average EER and 10.370% pooled EER, outperforming all evaluated continual-learning baselines. Across the five controlled protocols, RAMI yields the lowest common-average and pooled EER. Component ablations, limited-data experiments, and cross-backbone evaluations further validate the proposed design.