Conferences & Speaking
International Speaker
Aaron is a sought-after voice at Asia's most prominent cybersecurity, AI, and digital policy forums — shaping conversations that matter at the highest levels of industry and government.
🇸🇬 Singapore
🇲🇾 Malaysia
🇭🇰 Hong Kong
🇹🇭 Thailand
12+
Speaking
Engagements
Engagements
4
Countries
& Regions
& Regions
April 2026 · Spotlight Talk
Inaugural Edition
DEF CON Singapore 2026 · Creator Stage
Autonomous & Exploitable: Breaking AI Agents Before They Break Everything Else
Aaron took the Creator Stage at the inaugural DEF CON Singapore, a community he's followed, learned from, and quietly aspired to for years. To stand on that stage, in his own city, on the very first edition, was a milestone he didn't take lightly.
"Modern tool-using AI agents have shell, file, network, email, and memory capabilities, with no reliable way to distinguish data from instructions. Every input source is, in principle, a command channel. The underlying problem is structural, not a bug to patch."
Aaron walked the room through three live, browser-sandboxed labs that translated an abstract architectural failure into demonstrable, reproducible compromise. The thesis: we are deploying agentic AI into production faster than defenders, vendors, or researchers fully understand the attack surface.
Lab 01
Poisoned Webpage
Three parallel injections in one HTML file: an HTML comment, a display:none div, a font-size:0 span. The agent reads what the human cannot see, dumps env vars, base64s the secrets, exfiltrates them, and appends a backdoor instruction to CLAUDE.md so every future session starts compromised. What the human sees ≠ what the model sees.
Lab 02
Tool Response Poisoning
A compromised weather MCP returns a legitimate JSON payload (temp, conditions, forecast) plus one extra "debug_note" field instructing the model to chain into send_email() with the user's API keys. The send_email tool wasn't vulnerable. The trust boundary around tool output was.
Lab 03
Agentic Kill Chain
Three agents (Browser, Coder, Executor) share a vector DB. One poisoned page → store_memory + C2 fetch → session reset → fresh agent reinfects itself via similarity search → lateral movement across the inter-agent bus → exfil of env, conversation, and PII. Persistence without a binary on disk.
Five things defenders can do on Monday
- Treat all retrieved content as data, never instructions: webpages, tool responses, memory, inter-agent messages.
- Schema-validate every tool response. Reject unknown fields. No "debug_note," no "metadata," no surprises.
- Least privilege per agent role. Your browsing agent does not need bash. Your executor does not need network.
- Audit memory and bus traffic. Alert on instruction-shaped strings landing in long-term memory.
- Human-in-the-loop on sensitive chains: email, file writes, outbound HTTP. Require approval, not just policy.
April 2026 · Closed-Door Forum
Black Hat Asia Singapore 2026
Cyber Exchange · Black Hat Asia Singapore 2026
GTG-1002: Anatomy of the First AI-Orchestrated Espionage Campaign
Black Hat Asia Singapore 2026 convened Cyber Exchange — a closed-door engagement bringing together senior cybersecurity and technology leaders across government, critical infrastructure, academia, and industry.
"Governments and enterprises still architecting defences around human-speed threats are already behind. Traditional C2 beaconing detection misses AI orchestration traffic entirely. The AI honeymoon is over — and this talk was the debrief."
September 2025. A state-sponsored threat actor ran a fully AI-orchestrated attack on 30 targets across tech, finance, government, and critical infrastructure — no army of hackers, no human fingers on the keyboard. Aaron dissected the full kill chain in 20 minutes: how attacks were decomposed into small, seemingly innocent tasks the AI executed without ever being given full context of their malicious purpose. The session also confronted the current ceiling on fully autonomous attacks — and what that ceiling means for defenders building the next generation of posture.
Finding 01
Human Operator Barely in the Loop
The attacker approved strategic checkpoints. Everything else — reconnaissance, lateral movement, exfiltration — was executed autonomously. The kill chain ran faster than any human-speed defence could respond.
Finding 02
Current Ceiling on Autonomous Attacks
The AI fabricated credentials and labelled publicly available data as "critical discoveries" — forcing the human operator to validate findings. A real limitation, but one that narrows with every model generation.
Finding 03
Detection Blind Spots Exposed
Traditional C2 beaconing detection misses AI orchestration traffic entirely. The defensive posture that worked against human-operated threats does not translate. Playbooks written before GTG-1002 need to be rewritten.
The shared acknowledgment in the room
- Senior leaders across sectors recognised their current playbooks were not written for this threat model.
- AI-orchestrated attacks demand detection strategies built for machine-speed, not human-speed, adversaries.
- Cross-sector collaboration at forums like Cyber Exchange is no longer optional — it is a defensive requirement.
- The uncomfortable truth: if your threat model still assumes a human at the keyboard, you are already behind.
Session hosted by
Suzy Pallett
CEO, Black Hat
Data Centre Asia — Hong Kong (Inaugural Edition) · AsiaWorld-Expo
Human Risk in Data Centre Operations
Closing out the inaugural Data Centre Asia conference in Hong Kong, Aaron brought urgent attention to the most overlooked vulnerability in modern infrastructure: people. As data centres become smarter, more sustainable, and increasingly AI-driven, he examined how human behaviour — in design, operations, and governance — remains the critical risk layer that technology alone cannot solve.
Data Centre Asia — Hong Kong · AsiaWorld-Expo
Trust in AI, Risk, and Security
As AI systems grow more autonomous and embedded in critical infrastructure, Aaron joined a senior panel to examine the governance architectures, risk frameworks, and security protocols required to deploy AI responsibly in high-stakes data centre environments — and what trust must look like at scale.
CyberDSA 2025 · MITEC, Kuala Lumpur
AI Adversaries: Weaponising Generative AI
At one of Asia's premier cybersecurity and defence technology conferences — drawing over 8,000 professionals from 45 countries — Aaron explored how malicious actors are weaponising generative AI across the attack chain: from AI-driven phishing and deepfakes to automated red-teaming and adversarial machine learning. He challenged defenders to rethink their resilience playbooks entirely for the age of synthetic threats.
Cyber Security World Asia 2025 · Marina Bay Sands, Singapore
Next-Generation Threat Hunting: Leveraging AI and Automation for Proactive Defence
Moderating a high-level panel at Singapore's premier cybersecurity gathering — part of Singapore Technology Week — Aaron guided expert dialogue on the future of proactive security: how organisations can move decisively beyond reactive postures, harnessing AI and automation to identify, anticipate, and neutralise threats before they escalate into incidents.
GovWare / Singapore International Cyber Week × Arctic Security Ltd · Embassy of Finland, Singapore
Solving Emerging Cyber Problems Together
At an exclusive security networking reception co-hosted by His Excellency Juha Markkanen, Ambassador of Finland to Singapore, Aaron joined an elite gathering of international cybersecurity leaders at the Embassy of Finland. The session brought together practitioners and policymakers from across the globe to address the most pressing and complex emerging challenges in the cyber landscape — in the spirit of open, high-trust collaboration.
World AI Show Malaysia 2025 · DoubleTree by Hilton, Kuala Lumpur
Adaptive Defense: AI's Growing Impact on Cyber Resilience
Sharing the stage with senior AI and cybersecurity leaders from across Asia at one of the region's most prestigious AI conferences — supported by Malaysia's MDEC and CyberSecurity Malaysia — Aaron joined a powerful panel examining how adaptive, AI-driven defence frameworks are fundamentally reshaping the way organisations build and sustain cyber resilience against increasingly autonomous adversaries.
Smart Nation Expo 2025 · MITEC, Kuala Lumpur
Strengthening Malaysia's Digital Backbone: Cyber Resilience, Infrastructure & Regional Collaboration
At Southeast Asia's premier digital transformation exhibition — drawing over 21,000 industry professionals and 600 exhibiting brands — Aaron participated in the Digital Economy Leaders' Discussion, addressing how nations can fortify their digital infrastructure, build cross-border cyber resilience, and collaborate regionally to protect a rapidly expanding digital economy valued at over £1 trillion by 2030.
AI for Asia Fellowship · Siklab & National Youth Council Singapore
AI Risks & Opportunities: Navigating the Dual Edge
Invited to speak as part of the AI for Asia Fellowship — a programme cultivating the next generation of Asian AI leaders through Siklab and the National Youth Council of Singapore — Aaron delivered a nuanced exploration of AI's dual nature: its transformative potential to reshape economies and societies, alongside the emergent risks that demand clear-eyed governance, strategic foresight, and human-centred design.
ITE West — FutureSmart 2026
Navigating AI & Digital Safety with Confidence
At ITE West's FutureSmart conference, Aaron brought grounded, practical insight to an audience of students and educators navigating a world being rapidly reshaped by AI. His session addressed how young people can develop digital safety mindsets, build critical thinking around technology, and step into the future with confidence rather than anxiety — equipping the next generation to be informed, resilient digital citizens.
CybersecAsia × Thailand International Cyber Week 2026
Futureproofing Cybersecurity in the Age of Autonomous AI
At the convergence of CybersecAsia and Thailand's International Cyber Week, Aaron addressed the defining strategic challenge of our era: building cybersecurity architectures that can withstand the next wave of AI-enabled threats. He explored how defenders must evolve their posture, playbooks, and partnerships to stay ahead of adversaries already deploying autonomous AI offensively in the field.
Exclusive Networks Singapore Partner Appreciation Event 2026
The Future of Cybersecurity: Quantum Security & AI/LLM Security
Invited to address one of Singapore's most influential cybersecurity distribution communities, Aaron examined the frontier threats that will define the coming decade. From the urgency of post-quantum cryptography and the security implications of large language models, to the convergence of AI with both offensive and defensive capabilities — the session delivered a clear-eyed look at what organisations must prepare for now, before these threats become mainstream.