With the rapid advancement of computing hardware, learning algorithms, and the explosive growth of data, machine learning (ML) technologies have been widely adopted across diverse domains such as facial recognition, intelligent video surveillance, autonomous driving, and beyond. Despite their remarkable success, the security and privacy implications of ML systems remain insufficiently understood. In particular, adversarial machine learning continues to expose unique vulnerabilities in models, yet there is still a lack of systematic methods to assess and improve their robustness. At the same time, growing public awareness of data privacy raises critical concerns about how to train and deploy ML models without compromising sensitive information.

In addition, the rapid progress toward increasingly Artificial General Intelligence (AGI) systems (e.g., foundation models and AI agents) introduces new risks that go beyond conventional ML settings. These include issues of content provenance, detecting and mitigating mis/disinformation, ensuring trustworthy deployment at scale, and safeguarding both AGI and AI agents against misuse or adversarial manipulation.

This workshop aims to bring together researchers and practitioners to explore these pressing challenges. We solicit high-quality contributions addressing a wide range of topics, including but not limited to adversarial learning, robust algorithm design and evaluation, privacy-preserving machine learning techniques, and secure ML system deployment. The workshop will provide a forum for participants to exchange cutting-edge ideas, present novel solutions, and discuss emerging trends that bridge theoretical advances with real-world applications.

Submission Guidelines

All papers must be original and not simultaneously submitted to another journal or conference. Submissions must be anonymous, with no author names, affiliations, acknowledgement or obvious references. Each submission must begin with a title, short abstract, and a list of keywords. The introduction should summarize the contributions of the paper at a level appropriate for a non-specialist reader. All submissions must follow the original Springer LNCS format with a page limit of 20 pages (incl. references). Authors of accepted papers must guarantee that their paper will be presented at the conference and must make a full version of their paper available online. Submissions not meeting the submission guidelines risk rejection without consideration of their merits. It is strongly encouraged that submissions be processed in LaTeX.

Final Version Preparation (Camera-Ready)

Authors of accepted papers must prepare and submit a final (camera-ready) version according to the following requirements. Failure to comply with these requirements may result in exclusion from the workshop proceedings.

• Author Information: The final version must include the full author list, affiliations, and acknowledgements.
• Springer LNCS Template: The paper must strictly follow the official Springer LNCS format . The final version must not exceed 20 pages, including references.
• Consent-to-Publish Form: At least one author of each accepted paper must complete and sign the Springer Consent-to-Publish form . The completed form must be submitted together with the final version.

List of Topics

Topics of interest include, but are not limited to:

• ML for Security Applications --- software vulnerability detection, malware classification, spam/phishing detection
• Security Testing and Verification of ML Systems
• Adversarial Machine Learning --- attacks, defenses, and evaluation methodologies
• Robustness Analysis and Certified Defenses for ML Models
• Data Poisoning and Backdoor Attacks --- detection, mitigation, and resilience strategies
• Intrusion/Anomaly Detection/Prevention
• Privacy-Preserving ML --- scalable and privacy-enhancing techniques for ML such as differential privacy, secure multi-party computation, homomorphic encryption, and trusted hardware
• Watermarking, Ownership Verification, and Model Provenance
• Attacks and Defenses for Biometric Systems --- face, voice, and multimodal authentication
• ML for Program Analysis and Secure Software Engineering
• Emerging Threats in AI-Driven Systems --- autonomous systems, IoT, CPS
• Security of Large Language Models --- prompt injection, jailbreaks, and LLM-based security tools
• Security and Trustworthiness of AGI Systems --- content provenance, misinformation/disinformation detection, large-scale alignment and resilience
•  

Venue

The conference will be held in in conjunction with ACNS2027 (June 28 – July 1, 2027), Windsor Building, Royal Holloway, University of London, Egham, UK.

Contact

All questions about submissions should be emailed to simla2027.chairs@gmail.com.