Long papers
Up to 8 pages of main content, following ARR/ACL formatting
NLP/CORE Unranked
TrustAudio 2026
TrustAudio: Workshop on Secure, Safe, and Privacy-Preserving Speech and Audio AI
TrustAudio 2026 is a one-day workshop co-located with AACL-IJCNLP 2026 in Zhuhai, China, focused on advancing secure, safe, and privacy-preserving speech and audio AI. It brings together researchers to address risks across the audio AI lifecycle, including adversarial attacks, deepfakes, privacy leakage, and safety evaluation in multimodal and multilingual systems.
Key deadlines
Verified September 9, 2026Full paper
September 15, 2026
AoE
Workshop timeline
Submission and decisions
Full paperKey deadline
September 15, 2026 · AoE
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
Short papers
Up to 4 pages of main content, following ARR/ACL formatting
Research areas in scope
01
Topics of Interest
Digital adversarial attacks on speech and audio models, including ASR, TTS, and speaker recognitionPhysical-world and real-world audio attacks, including over-the-air and replay attacksRobustness, safety, and reliability evaluation of audio-language and multimodal models, benchmarks, and protocolsDeepfake and synthetic voice misuse, including voice cloning and impersonationHallucinations, misalignment, and other unintended behaviors in audio and multimodal generative modelsData poisoning and backdoor attacks across the model lifecycle, including data, training, fine-tuning, and releasePrivacy risks and privacy-preserving training or deployment for speech and audio systemsModel extraction and IP attacks, plus provenance and attribution for synthetic audioDefensive methods and system-level safeguards for speech, audio, and multimodal systemsSafety and robustness in multilingual, multicultural, and underrepresented languagesSafety evaluation, red-teaming, and alignment of LLMs, audio-language models, and multimodal foundation modelsPrompt injection, jailbreaks, instruction-following failures, and unsafe tool use in LLMs and multimodal frameworksUncertainty, calibration, refusal behavior, and responsible generation across language, speech, audio, and multimodal systems
Policies worth checking twice
- Submissions must be anonymized for double-blind peer review
- Papers must follow ARR/ACL formatting guidelines
- Cross-submissions and dual submissions are not allowed; work must be original and not under review elsewhere
- Submissions must be made via OpenReview
- ARR-committed papers may be submitted via the OpenReview ARR Commitment site if compliant
- Papers must include required ethics sections as per ARR requirements
- Accepted papers are expected to be presented in person at the workshop
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026