The HTxAI movement

With its mission of advancing science and technology, force multiplying the Home Team, and securing Singapore’s future, HTX has been applying cutting-edge developments in Artificial Intelligence (AI) to public safety and security

With advancements in machine learning and deep learning, which have laid the foundations for Generative AI, HTX’s AI engineers are harnessing Generative AI’s powerful capabilities to build innovative AI solutions to enhance homeland security and improve the Home Team’s citizen services.

Together, these constitute the HTxAI movement, which was announced on 1 June 2024. (“HTxAI” is the abbreviation for “Home Team multiplied by AI”.) The agency-wide HTxAI movement comprises:

AI Central

The AI Central division manages, tracks, and supports overall AI functions across HTX. It consists of the following 3 sub-areas:

• AI Masterplanning, Governance & Collaboration: Manage, track and support overall AI functions across HTX; develop and ensure AI Governance regulatory & standards compliance; ecosystem outreach, partnerships & collaborations
  
• AI Solutions & Design: Engage HTD stakeholders, facilitate service design workshops, and ensure proposed solutions meet objectives and user needs
  
• AI Training & Solutions Certification: Develop and implement AI competency frameworks, training programmes, certification processes, and assessment standards for AI and digital initiatives

AI Enablers

The HTxAI movement also includes the following 3 enablers:

• AI Platform: Enhance developer experience; develop, operate, and maintain secure DevSecOps pipelines and productivity tools for continuous AI development in classified environments
  
• AI Infrastructure: Design and build a robust and scalable AI compute infrastructure for classified applications, accelerating Home Team’s development and adoption of AI including compute-intensive Gen-AI applications
  
• AI Data Engineering: Design and implement the data infrastructure, including data connectors and pipelines that support AI model development and deployment, ensuring data governance, quality, and security

AI Strategy

The HTxAI movement is aligned with Singapore’s National AI Strategy (NAIS 2.0) to position Singapore as a global leader in AI by 2030. As part of its AI strategy, the HTxAI movement will leverage partnerships and collaborations with industry and academia to strengthen HTX’s internal AI capabilities, and advance its AI research and development (R&D).

AI R&D

The HTxAI movement’s R&D areas include:

• Multi-modal AI & Agentic AI: Seamless integration of language, vision, and other modalities for comprehensive understanding and situational awareness, and enhanced user interactions for more comprehensive understanding. AI Agents can act autonomously and effectively interact with their environment with enhanced factual consistency and relevance of generated content.
  
• Language AI: Speech-to-text capabilities, enhanced language understanding (including nuances, slang and context) and broader language support to improve global accessibility and inclusivity
  
• Embodied AI: Advanced Robotics Capabilities such as navigation using AI Vision, and Human-Robot Collaboration through natural language conversations, to augment human in the three Ds - Dangerous, Dull and Dirty tasks
  
• Vision AI: Improved accuracy in image and video processing, precision of object detection, recognition, and classification and vision-AI techniques such as deep-fake detection
  
• Emerging AI: Accelerate innovative technologies, applications and opportunities for AI in Homeland Security & Enforcement
  
• AI Safety & Security: Advanced defences and detection to safeguard against adversarial attacks and manipulations

To enable and accelerate AI R&D initiatives, relevant and high-quality data is needed to power the training and development of advanced AI models like LLMs and large multi-modal models (LMMs). Given the potential sensitive nature of such data, safeguards and measures have been put in place to ensure the quality, security, and privacy of data. This includes the implementation of data governance controls for to oversee responsible data collection, management and utilisation.

AI Infrastructure

Additionally, HTX is also investing in state-of-the-art graphics processing unit (GPU) infrastructure to allow its AI models, LLMs, and LMMs to handle data streams such as video footage from surveillance or body worn cameras, and to process these large amounts of data quickly and efficiently.

Furthermore, the implementation of DevSecOps (development security operations) and MLOps (machine learning operations) is a crucial part of the HTxAI movement. These pipelines enable secure and seamless collaboration between product teams, data scientists, AI engineers, and platform and security teams. In addition, the integrated automated workflows, continuous integration and delivery (CI/CD) processes, and robust security controls, ensure the security, reliability and scalability of AI systems.

AI Governance

Given AI’s potential to significantly influence society, The HTxAI movement is committed to ensuring the responsible and ethical development and deployment of its AI systems. HTX’s AI Central team is crafting a comprehensive Home Team AI governance playbook for the HTxAI movement, which contains operational guidelines and procedures for AI development, deployment, and usage.

The HTxAI movement’s AI governance structure and rules  will ensure transparency, explainability, and accountability in its AI systems. The movement’s responsible AI approach will be guided by human-centred design principles, emphasising inclusivity, diversity, and fairness, and prioritising the well-being and safety of individuals and society as a whole.