Hon Hai Research Institute at COMPUTEX 2026: Enterprise AI vs. AI Cybersecurity

At the COMPUTEX 2026 Pre-Show Forum, Wei-Bin Lee, CEO of Hon Hai Research Institute, gave a keynote titled "Enterprise AI vs. AI Cybersecurity.” Wei-Bin Lee explores the challenges of securing AI-powered organizations, from shadow AI and AI agents to prompt injection attacks and governance. With the proliferation of Agentic AI, traditional security approaches are no longer enough – enterprises must learn to balance innovation with accountability and cybersecurity. Cybersecurity threats have shifted from malicious code to natural language attacks.

Through techniques like prompt injection, hackers can now manipulate internal AI systems into writing malware, compromising mail servers, and breaching databases without traditional coding tools. Additionally, AI tools have compressed the time hackers need to exploit system vulnerabilities to mere minutes or hours, making accelerated defense measures imperative.

Wei-Bin Lee stressed that the era of Generative AI demands an overhaul in cybersecurity strategy, arguing that threats which spread like airborne diseases cannot be thwarted with traditional, rigid defense systems. Enterprise defense strategies must demand rigorous vetting to mitigate risks. Vendors should have to provide clarity regarding their AI models, including on the training methodologies, the data sourcing, the security measures, and the level of interpretability.

Because AI is no longer bound by predefined logic, it must be governed with the same accountability we apply to humans. Simply put, AI may 'know' the rules without choosing to follow them. Consequently, when an AI’s actions have irreversible impacts, 'Human-in-the-loop' oversight is non-negotiable. Enterprises must establish a robust, multi-tiered framework spanning legal, operational, and technical accountability.

Watch the full keynote on @COMPUTEXTAIPEIshow YouTube: https://youtu.be/H2Y2tDhDoq0?si=MCK0A6IzHejCbkfP