AI is moving faster than most security teams can keep up with. As enterprises rush into Generative AI and Agentic systems, the risks are piling up, and many leaders still have no clear plan to manage them.

In this episode of Decoding the Future, Stanley Tsang from the Cybersecurity Agency of Singapore (CSA) and Dr. Chris Marshall from IDC break down the real-world threats that come with scaling AI across the enterprise. From LLM vulnerabilities and guardrail bypasses to data sovereignty and AI-enabled cyberattacks, this conversation reveals what companies must fix before AI becomes unmanageable.

===

00:00 Introduction
01:36 Open Source AI Risks & Scaling Challenges
02:28 Why open source LLMs break down in production
03:34 CSA’s view: security must start on day one
04:52 Agentic AI and the expanding attack surface
06:11 How different industries are adopting AI
07:14 Using AI to strengthen cybersecurity operations
07:44 Deepfakes, phishing, and AI-enabled threats
08:17 Are enterprises prepared for secure AI design?
09:07 How LLMs expose sensitive data
11:36 The truth about guardrails and bypassing them
12:28 Risks of connecting LLMs to real workflows
12:59 From shadow AI to AI Centers of Excellence
18:18 Who owns AI security: CTO, CISO, or CEO?
22:42 The reality of data sovereignty and fragmented regulations
27:30 Solving the sovereignty issue with private on-prem models
29:41 Key takeaways for enterprise leaders

Thank you for listening!

Discover more content like this on Decoding the Future: https://bit.ly/484I3Zb
Learn more about Fujitsu's AI Solutions - https://mkt-asia.global.fujitsu.com/d...