📄 Description
This workshop, titled "DDL 2.0: AIGC Safety from Dual Perspectives of Model and Content", is a follow-up to our successfully organized IJCAI 2025 event "DDL1.0: Deepfake Detection, Localization, and Interpretability". We propose to host this workshop at IJCAI-ECAI 2026 to address the emerging challenges at the intersection of AI safety.
The rapid development of AI-generated content (AIGC) technologies has created an urgent need for comprehensive security frameworks. Model-level security defenses serve as the "first line of defense", aiming to block malicious generation at the source, while content-level identification functions as the "last line of defense" after risk propagation, used to detect and curb already-generated harmful content.
The primary goal of this workshop is to break down barriers across modalities and tasks, enabling end-to-end collaborative analysis from generative models to emitted content, and to build a systematic security framework that spans the full lifecycle of AIGC. By bringing together diverse researchers to discuss these critical issues, we aim to accelerate progress in AIGC safety technologies and catalyze transformative solutions.
Aligned with IJCAI-ECAI's mission of propelling AI innovation, this workshop provides a pivotal platform for impactful discussions and prospective solutions to pressing AI safety issues. The workshop will feature invited talks, paper presentations, and competition results.
Key Topics Include:
- Cross-modal AIGC detection and unified anti-forgery frameworks
- Safety alignment of generative models and input defense mechanisms
- Multicausal provenance and accountability for AIGC
- Security evaluation and red-team benchmarking for generative models
- AIGC forgery localization and interpretability analysis
- End-to-end AIGC security and defense systems
📄 Call for Papers
We call for submissions aligned with this workshop's themes. Submissions will be evaluated, and selected works will be presented as poster papers, oral presentations, or the best paper. Awards will be conferred, and the chosen papers will be showcased during the final workshop event. Submissions may be up to 7 pages long and adhere to the formatting guidelines specified for IJCAI-ECAI 2026 (main track). Three members of the Program Committee will review each submission.
Keynote Speakers
