In today’s AI-powered world, it’s not just about adopting innovation, it’s about securing it. As large language models (LLMs) become more integrated into business operations, a new threat is emerging, LLM Jacking, a form of manipulation that compromises AI tools and exposes sensitive data. 

This isn’t a future concern, it’s happening now. And understanding how to respond is crucial. 

In the inaugural episode of The High-Growth Equation, a new video series from TD SYNNEX, we explore the rapidly evolving threat landscape of LLMs, and how partners can proactively protect their customers. 

Lawrence Roberts with TD SYNNEX, host of the series, sat down with Matt Bingaman, Security Solutions Architect and AI Security Champion at TD SYNNEX, to demystify two critical security topics, LLM jacking and LLM jailbreaking, and offer clear strategies to defend against them. 

What Is LLM Jacking? 

LLM Jacking refers to an attack where a malicious actor manipulates a large language model into delivering inaccurate or harmful results, often without altering the model itself. These attacks exploit the underlying prompt or data flow, essentially “hijacking” the model’s outputs for unintended use. 

The stakes are high. As organizations integrate AI across customer service, productivity tools, and sensitive business functions, LLM jacking can expose private data or compromise trust in AI-enabled systems. 

To help you navigate this critical topic, we’ve outlined key moments from the discussion. The following time stamps walk you through essential insights and practical strategies: 

LLM Jacking vs Jailbreaking (0:45 – 2:02) 

While both LLM Jacking and Jailbreaking involve manipulating model behavior, the key difference is intent and method. 

• LLM Jacking corrupts the flow of legitimate prompts to twist outputs for malicious purposes. 
• LLM Jailbreaking, on the other hand, refers to bypassing the guardrails or safety restrictions placed on a model, prompting it to generate responses that would otherwise be blocked. 

Understanding both threats is key to building comprehensive security strategies. 

Best Practices for Securing LLMs (2:02 – 4:40) 

Defending against these threats begins with strong AI governance. Some recommended practices include: 

• Implementing prompt validation and monitoring. 
• Maintaining a human-in-the-loop review process. 
• Using zero-trust architecture principles for AI pipelines. 
• Applying data input/output filtering. 
• Routinely testing and auditing AI behavior. 

As Bingaman explains, “It’s all about intentionality, AI should serve your goals, not the other way around.” 

The Importance of Data Protection in AI (4:41 – 6:26) 

Since LLMs often train or adapt based on user data, unguarded information can become a vulnerability. Protecting that data, both in training and in application, is essential to prevent exploitation. 

Partners and businesses should work toward: 

• Data minimization, using only what’s necessary. 
• Anonymization or pseudonymization of sensitive information. 
• Ensuring compliance with regional data protection laws. 

Future Trends in LLM Security Threats (6:27 – 8:09) 

Looking ahead, LLM-targeted threats will continue to evolve. We can expect: 

• More automated, AI-driven attacks. 
• Threat actors exploiting multi-modal models (text, voice, and images). 
• A growing need for AI red-teaming (simulated attacks) and continuous monitoring to safeguard systems. 

As Matt shares, “The threat surface is dynamic, but so is our ability to adapt.” 

How Partners Can Get Support (8:11 – 9:06) 

TD SYNNEX offers a wide range of resources to help partners build secure AI practices, including: 

• Security advisory services. 
• Training and enablement programs. 
• Partner tools to help assess and mitigate risk. 

Sharing relevant threat intelligence and practical solutions with partners helps ensure AI continues to fuel progress while minimizing risks. 

Explore More in The High-Growth Equation 

This article is part of The High-Growth Equation, designed to help you succeed in an industry that’s constantly evolving. Here, you’ll find clear insights to cut through complexity, innovative ideas that create new opportunities, and actionable strategies to turn challenges into growth. 

We focus on the most powerful, high-growth technologies shaping your business, from hybrid cloud and AI to cybersecurity, data, and IoT, providing the expert guidance you need to stay ahead and make the most of every opportunity. 

Stay tuned for more insights from TD SYNNEX leaders and co-workers around the globe by following us on LinkedIn, Instagram, and on our Newsroom.