Artificial intelligence (AI) continues to redefine how organizations operate while simultaneously changing the field of data security. However, this shift brings both opportunities and challenges. AI can bolster security defenses and streamline operations. Yet, it also amplifies risks, providing attackers with more sophisticated tools to exploit vulnerabilities.

In this episode of the #shifthappens podcast, Bob Buda, President of Buda Consulting, shares his insights on how clean, well-managed data is the cornerstone of secure and successful AI adoption. He also offers actionable strategies for business leaders seeking to harness its potential and discusses his vision for how AI can drive meaningful societal change.

AI: A Multiplier in Data Security

AI acts as a force multiplier in data security. For defenders, AI-driven tools can detect anomalies, predict threats, and automate responses faster than ever. However, attackers also leverage AI to craft more convincing phishing attacks, identify system weaknesses, and exploit vulnerabilities at scale.

“The fundamentals of data security have not changed. What's changed is the ability of the bad guys to take advantage of the vulnerabilities that are there. AI is a multiplier — both for security teams and attackers,” Bob explains.

The critical insight? AI doesn’t replace security fundamentals; it amplifies their importance. Without robust security practices, organizations may inadvertently give AI-powered adversaries an advantage. Leaders must, therefore, prioritize foundational security measures and recognize that AI’s effectiveness depends on the quality and integrity of the data it consumes.

For managed service providers (MSPs), leveraging data analysis is key to unlocking AI’s full potential. Understanding the difference between outcomes (desired results) and drivers (factors influencing outcomes) is essential for AI adoption. Just as excelling in a race is an outcome while running faster is the driver, optimizing structured data leads to better AI performance.

Fundamentals of Data Security: The Cornerstones of AI Success

Bob identifies three essential data security principles that form the bedrock for secure AI adoption:

1. Multi-Factor Authentication (MFA): MFA is one of the most cost-effective and impactful measures organizations can deploy to protect against breaches. Relying on passwords alone is no longer sufficient when AI enables increasingly sophisticated attacks. MFA adds an essential layer of security, ensuring that even if credentials are compromised, unauthorized access remains unlikely.

Bob highlights this point by saying it is a “low-hanging fruit.” It is a simple, cost-effective solution that significantly reduces the risk of unauthorized access that every organization should have in place.

Leadership takeaway: Ensure MFA is implemented across all critical systems, especially those involving sensitive data and AI-driven processes.

2. User Training: People remain the most significant vulnerability in any security framework. AI-powered phishing attacks are now more convincing and harder to detect. User training, therefore, is critical. Employees must understand how to recognize and respond to AI-generated threats.

Bob notes, “Phishing schemes are getting better. They used to have some type of misspelling or a clue that it’s not good. But with AI, they can build sophisticated ones that can fool the best of us.”

Leadership takeaway: Invest in regular, AI-specific security training. Equip teams with the knowledge to identify deepfake content, AI-generated phishing attempts, and other sophisticated threats.

3. User Access Control: Properly managing user access ensures sensitive information is only available to those who truly need it. Unrestricted access can lead to accidental data leaks or provide attackers with easier targets.

Bob illustrates this with an analogy comparing it to a house where all valuables are stored in a safe inside a locked bedroom. “That’s what many companies do. They’re protecting the perimeter but paying very little attention to the database itself. So, if somebody manages to get through, everything is more vulnerable than it needs to be,” he adds.

Leadership takeaway: Adopt a zero-trust approach. Implement strict access controls and regularly review permissions to ensure they remain appropriate.

Structured vs. Unstructured Data: Why It Matters for AI

AI’s performance is deeply influenced by the nature of the data it processes. Bob differentiates between two primary data types:

• Structured Data: Structured data, stored in well-organized formats like relational databases, is easier for AI to process. This data is labeled, categorized, and ready for immediate analysis, enabling AI models to deliver faster, more accurate insights.
• Unstructured Data: Unstructured data – emails, documents, audio files – lacks a predefined format. While AI can process unstructured data, doing so requires additional resources. This addition can lead to higher processing costs and lower accuracy.

Bob predicts a trend where more unstructured data will transition into structured environments. With proper data preparation, he states, AI can be utilized more effectively.

From Chaos to Clarity: Preparing Data for Secure AI

AI is only as effective as the data it learns from. To build a strong, secure foundation for it, organizations must take deliberate steps to ensure data readiness:

1. Label everything right: Messy data leads to messy AI outcomes. While it can handle unstructured data, well-labeled datasets dramatically improve accuracy performance. Clear classification reduces processing costs, enhances model accuracy, and accelerates time to insight.
2. Get to know your data: You can’t protect what you don’t understand. Conduct a comprehensive data inventory and identify all data sources, locations, and uses. Breaking down data silos enhances security and ensures AI models work with complete, accurate datasets.
3. Secure the core: Protect your database first before AI starts making decisions. Strengthen security at the data level to prevent breaches and ensure that insights are trustworthy.
4. Use AI to prep for AI: AI-powered threats are evolving; your team should too. Data quality issues, such as duplicates, inaccuracies, and inconsistencies, can derail AI projects. Use general AI models for data cleansing before training specialized models to establish accurate and meaningful relationships between data points.

For growing MSPs, structuring and managing data effectively requires the right expertise. Key roles such as data analysts, AI specialists, and cross-functional teams should collaborate to maintain high-quality, secure data for smarter AI adoption.

Building a Secure AI Future Starts with Clean Data

AI’s promise is vast, but its success hinges on one critical factor: clean, secure data. Leaders who prioritize data governance, invest in foundational security measures, and embrace AI’s transformative potential will be best positioned to navigate the evolving digital landscape.

Bob’s insights serve as a roadmap for organizations seeking to leverage AI responsibly. By understanding the role of clean data, addressing security fundamentals, and preparing data ecosystems for AI integration, business leaders can unlock new opportunities while mitigating risks.

Beyond business, Bob envisions AI and data making a meaningful impact on healthcare. He shared his dream of using AI to personalize nutrition based on an individual's genetic makeup, aiming to prevent diseases rather than just treating them.

“What we eat has a significant impact on our health. What I'd like to see is AI used to pinpoint the kind of food an individual needs to eat to prevent disease rather than to treat it later. So, that’s what I’d love to see. It’s something I’d like to work on,” he explains. His vision showcases AI’s potential beyond security and business efficiency; it opens doors to improving quality of life through personalized, data-driven health solutions.