AIFT and SYSTEX, with AWS as co-host, convened industry leaders on September 5 to discuss how to balance AI innovation, compliance, and security. Nearly 40 industry leaders from finance, high-tech manufacturing, retail, and the public sector participated in the discussion.
Deloitte: The Evolution from GenAI to Agentic and Physical AI
Hank Hsieh, Senior Manager at Deloitte framed AI’s trajectory—from GenAI, through Agentic AI, to Physical AI—echoing NVIDIA’s vision of a dual workforce, both biological and digital. Yet, he cautioned that enterprises often focus only on visible tools like ChatGPT, overlooking critical foundations: compliance, privacy, security, ethics, and a robust implementation framework.
SYSTEX: AI Agents and New Business Ecosystems
Duncan Tu, Assistant Vice President at SYSTEX highlighted the shift from traditional SaaS to AI Agent as a Service (AIaaS)—ushering in new B2AI2B and C2B2AI2B2C business models. He posed the vital question: as AI adoption accelerates, how can zero trust architectures adapt, and how do we guard against AI hallucinations? The answer lies in mature governance and trusted AI foundations.
AIFT: GenAI Model Vulnerabilities and Defense Frameworks by Vulcan
Kevin Ke, Senior AI Engineer at AIFT shared Vulcan’s latest research efforts in identifying vulnerabilities among leading GenAI models—such as prompt injection, data leakage, supply chain attacks, and data poisoning. He stressed the importance of multi-layered defenses—like input/output validation, red teaming, vulnerability scanning, and multilingual testing—supported by compliance and zero trust for responsible AI deployment.
AWS: Amazon Bedrock’s Enterprise-Ready Architecture
Tim Yeh, Security Specialist SA at AWS introduced Amazon Bedrock, spotlighting its AgentCore (rapid deployment of AI agents) and Guardrail (real-time security controls). Bedrock’s architecture supports zero trust with encryption, IAM integration, observability, and audit capability—making secure, compliant Agentic AI deployment feasible across sectors.
AIFT: Vulcan Attack & Protect — Dual Defense in Practice
Frank Liao, Director of Partnership at AIFT, showcased Vulcan Attack, which automates red-teaming to detect vulnerabilities—shrinking testing time from 200 to just 3 hours—and Vulcan Protect, offering real-time monitoring, policy-based safeguards, and flexible deployment. Recognized by the OWASP AI Exchange, Vulcan continues collaborating with global leaders to reinforce GenAI security and compliance.
Conclusion: Bridging Innovation with Compliance
As GenAI and Agentic AI embed deeper into enterprises, balancing innovation with governance and security becomes indispensable. This event offered a holistic blueprint—from regulatory frameworks and governance models to cloud-native strategies and hands-on testing—that empowers organizations to adopt AI confidently.
AIFT is committed to building a responsible, secure, and trustworthy GenAI adoption with partners including AWS, SYSTEX and Deloitte in Taiwan and beyond.
