Advances in Trustworthy AI: Secure Intelligent Systems

Dear Colleagues,

As artificial intelligence (AI) technologies continue to be deeply integrated into social, economic, and daily life, the trustworthiness of AI systems has become a central concern for both academia and industry communities. Although modern AI has demonstrated remarkable potential, it still faces critical challenges in terms of robustness, interpretability, fairness, and privacy preservation. Due to the high complexity and opacity of deep neural network models, the detection, analysis, and mitigation of these risks remain highly non-trivial. Consequently, building secure and reliable AI systems has emerged as a fundamental objective and a major research frontier.

This Special Issue focuses on the design of trustworthy AI models, with particular emphasis on robustness, interpretability, fairness, and privacy in modern AI paradigms. We aim to systematically investigate the key challenges in developing trustworthy AI systems, including, but not limited to, training data leakage, data and model bias, adversarial robustness vulnerabilities, model extraction and intellectual property infringement, the generation of false or harmful content, privacy inference attacks, the rapid evolution of sample forgery techniques, the high complexity of end-to-end multimodal models, and shortcut learning in deep neural networks.

Topics of Interest

We cordially invite submissions of high-quality original submissions on topics including, but not limited to, the following:

• Attack and defense technologies for AI systems

Methods for identifying and mitigating security threats to AI models, including adversarial attacks, backdoor attacks, data poisoning, and corresponding defense mechanisms.

• Explainable Artificial Intelligence (XAI) in deep learning

Techniques to enhance the transparency and interpretability of deep neural networks, providing human-understandable explanations for complex model predictions.

• Fairness, bias, and discrimination in AI systems

Approaches for measuring and mitigating biases in data and models, as well as ensuring equitable decision-making across different demographic groups.

• Privacy and security in federated learning

Secure and privacy-preserving learning in decentralized environments, including robust aggregation, differential privacy, and defenses against malicious clients.

• Model intellectual property protection and provenance tracking

Techniques for protecting model ownership and usage rights, such as watermarking, fingerprinting, and tracing mechanisms for verifying the origin and integrity of trained models.

• Detection of forged and synthetic samples

Methods for identifying manually generated, manipulated, or tampered data, including deepfake detection, synthetic data forensics, and dataset integrity verification.

• Real-world applications of AI in high-risk domains

Empirical studies and deployed systems demonstrating how trustworthy AI techniques can be applied in safety-critical settings such as healthcare and drug analysis, where errors, bias, or lack of transparency may lead to severe consequences.

• Trustworthiness of large language models

Methods for conducting security and privacy attacks on speech foundation models to expose vulnerabilities in representation learning, semantic understanding, and speaker modeling, thereby undermining model reliability and privacy protection.

Prof. Dr. Yun Li
Guest Editor

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the Special Issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 250 words) can be sent to the Editorial Office for assessment.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-anonymized peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Electronics is an international peer-reviewed open access semimonthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2400 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.