In this episode host Oracle Academy Program Manager Benelux & DACH Rik Bleeker speaks with Oracle expert Thomas Minne on the crucial intersection of Artificial Intelligence and data security, explaining why the Oracle Database is foundational to both. Drawing on his 15 years as a DBA and current role in Oracle's EMEA CTO office, Thomas shares practical strategies for implementing AI responsibly without compromising sensitive information. Listeners will learn how features like Real Application Security and Label Security safeguard data, why AI vectors require the same protection as raw data, and hear real-world stories from industries like finance and healthcare. Thomas also discusses how AI can both enhance and threaten security, and offers career advice to students keen on mastering IT fundamentals and safely experimenting with Oracle LiveLabs. Whether you're a student, educator, or just curious about AI security, this episode is packed with insights to help you innovate confidently and securely.

--------------------------------------------------

Episode Transcript:

00:00:00:00 - 00:00:35:13 Unknown Welcome to the Oracle Academy Tech Chat. This podcast provides educators and students in-depth discussions with thought leaders around computer science, cloud technologies, and software design to help students on their journey to becoming industry ready technology leaders of the future. Let's get started. Welcome to the Oracle Academy Tech Chat, where we discuss how Oracle Academy prepares the next generation workforce.

00:00:35:13 - 00:01:08:08 Unknown I'm your host, Rick Bleeker, are being the Oracle Academy program manager for the Benelux and the region. In this episode, Oracle expert Thomas Minne explores how artificial intelligence and data security intersect. And why'd Oracle Database is at the center of this? All with 15 years of DBA experience and now part of Oracle's EMEA CTO office. Thomas shares real world insights on how organizations can use AI responsibly while protecting sensitive data.

00:01:08:10 - 00:01:35:18 Unknown From healthcare to finance, I offers huge opportunities, but only if security is in from the start. The topics we're going to cover today is why I makes data security more important than ever. How Oracle databases features like real application security and label security protects sensitive data. The risk of vectors in AI workflows. Why they need the same protection as raw data.

00:01:36:00 - 00:02:12:12 Unknown Real world stories of securing large financial systems. AI as a double edge sword basically strengthening the defenses but also usable by attackers. Career advice for students master it fundamentals. Explore AI workflows, and practice safely with Oracle Life Labs. So security also today isn't a barrier. It's an enabler. With the right governments, access controls, and hands on learning. Students and educators can explore AI confidentially and responsibly.

00:02:12:14 - 00:02:45:19 Unknown Today, we're joined by Thomas Minne, Oracle database Expert and Data Security Blackbelt. Thomas, welcome and an absolute pleasure to have you on this podcast. As a start, Thomas, could you tell us about your background and current role at Oracle? Thank you Rich. Hello everyone. Well, I work in the EMEA office of the CTO at Oracle. So focusing on data security, I have spent 15 years as an Oracle DBA, starting in infrastructure and gradually moving closer to application development, helping teams build performance and world design systems.

00:02:45:20 - 00:03:16:20 Unknown I'm also an Oracle Certified Master. In my current role, I advise customers on protecting their most valuable assets and their data. I specialize in Oracle database security features like encryption, auditing, identity and access management, data masking also, and other security goodies. So my work combines engineering, technical representation and also collaboration with project management, all with one goal keeping data safe while enabling innovation.

00:03:16:23 - 00:03:41:11 Unknown That sounds good Thomas. Thank you so much. So today's topic is AI and security and domains. If I look at the top 60 global technical roles, I and cloud roles are strongly represented in the top ten in various roles and jobs. Why is security so important in the context of AI and how does it connect to the Oracle database?

00:03:41:13 - 00:04:15:12 Unknown Great question. First, AI isn't magic. It's just a set of tools that process data. And whenever we talk about data, security becomes essential. AI system use techniques like larger language models, lens vectorization and retrieval, augmented generation, and drag. All of these interact with data and sometimes highly sensitive data. For example, in healthcare, Oracle Database Store patient records. You can't connect that kind of database directly to a public AI model.

00:04:15:12 - 00:04:54:06 Unknown You could accidentally expose private medical information. I can absolutely add doctors and researchers, but only if access is controlled, the right person at the right time, and the right context. Without those safeguards, the risks outweigh the benefits. And that's the challenge. Making sure that even if the data are highly sensitive, we manipulate those data safely to sarcomas. Listening to the security on its own in it is already a challenge to what is your opinion, a security, an extra challenge for companies when enabling with new AI capabilities?

00:04:54:08 - 00:05:19:14 Unknown Exactly. Companies are no longer asking should we use AI? They are now asking how can we use AI securely? That's where Oracle Security Features comes in, helping organizations build secure, compliant pipelines for AI solution while still benefiting from AI. Spoiler. And is that challenge is not taken seriously. It will prevent them to move on with their project.

00:05:19:16 - 00:05:55:00 Unknown Got it. Thomas. So, as we all know, security is crucial for organizations. Community security processes, however, are often seen as painful or complex, time consuming and even slowing down projects. How do you look at this? It's true. Unfortunately, and security can seem like extra work, but it is the reality. You can't escape it. The potential damage from the breach to both individuals and companies is huge, and companies must also be compliant with the different regulations.

00:05:55:02 - 00:06:19:03 Unknown I like to flip the perspective. Security can actually. Oh, to move you to move faster. If you can guarantee that only authorized holes in your application can see sensitive data. You unlock the ability to use that data in AI workflows with confidence. Without that, companies tend to lock their data in a bunker safe, but effectively use less.

00:06:19:05 - 00:06:45:08 Unknown So security becomes an enabler. It's the condition that allows you to use AI responsibly. The closer AI tools gets to the data, the more fine grained access control becomes non-negotiable. Think of it as an investment. If you understand your data, its sensitivity, and how to control access, you can safely leverage powerful AI capabilities. Okay, okay, it almost got it.

00:06:45:08 - 00:07:09:18 Unknown But even then, still, there are multiple options. So why the Oracle database? There's a lot of new open source tools that could help build AI pipelines, aren't there? Yeah. It's true that there are two key points. First, the sensitive data there are already in the Oracle database. And many of our customers are already running critical system on our infrastructure.

00:07:09:19 - 00:07:41:22 Unknown And they're happy with the performance but also the availability of our platforms. So moving data into another system like an open source vector database forced you to duplicate security measures across platforms, adding complexity and risk. In addition, you may lose the benefits of a highly available and performant platform. So, Gunns Oracle Converged Database strategy lets you store data, owns and run multiple workloads SQL, graph, vector search, analytics, spatial all in one place.

00:07:41:22 - 00:08:13:21 Unknown So security is built into the core, not bolting on later. With Oracle 23 AI database, these capabilities are unified, enabling innovation without compromising protection. Okay, clear. Understood. Thomas, thanks for elaborating. And but how do you look then at the relationship between data governance on one side and AI security evolving on the other? Data governance is absolutely critical. The team help us understand what data we have.

00:08:13:21 - 00:08:40:08 Unknown It's sensitivity and applicable regulations. They're the bridge between technical teams and compliance requirements. Simply put, you can't design an effective security strategy if you don't know which data is sensitive. For example, when I worked at a major bank in Luxembourg, I collaborated with colleagues from the data governance and business intelligence team to design security for a new data warehouse.

00:08:40:13 - 00:09:18:22 Unknown We classified the data and determine how to secure it. Access was controlled along multiple axis data domain, so over 50 domains with different sensitivity level inside each subsidiary user and access only to certain subsidiaries data on play data only HR could access it and also customer personal information, GDPR compliance and protection of personal information. All these control were unforced directly in the database, so each user only saw what they were authorized to access for their workflow.

00:09:18:22 - 00:09:56:14 Unknown The same principle apply AI solution should only access allowed data features like Rear Application security, virtual Private database or label security can enforce these rules and ensures AI tools operate safely and compliantly. LeClair did. That explains about the relationship and Thomas could AI itself help improve database security in the future? You think absolute. For example, SQL server lets you run queries and generate reports on an Oracle database using natural language.

00:09:56:16 - 00:10:25:17 Unknown You can even use it for security testing. Ask the AI to find sensitive data and it can quickly produce a detailed report. This is just one example. I'm convinced we will see many ways AI tools can proactively increase database security, identifying risk, monitoring access, and even suggesting improvement. Wow, that that is fascinating, Thomas. But doesn't that also mean attackers could use the same AI tools to find former abilities faster?

00:10:25:17 - 00:10:51:12 Unknown And how do we defend against that? Exactly. It's a double edged sword. But using AI internally helps us to find weakness before attackers do. Security isn't a simple on off switch. It's the combination of multiple tools, proper configuration, and best practices that elevate your overall security posture. I advise our student to check what is the Oracle Maximum Security Architect here.

00:10:51:14 - 00:11:21:04 Unknown It will definitely help them to understand that it's the combination of many different building blocks that create a good security posture. Clear, clear. And is there an area of AI security that is often underestimated in European. Yes. Vectors. In AI, a vector is a numerical representation of data like text or image transform into a series of numbers used later for similarity search.

00:11:21:05 - 00:11:47:20 Unknown Think of it like a fingerprint of the data. Even without the original file, it can reveal information if not protected. Some assume vectors are harmless because there are just numbers, but sensitive data must be secured at the vector level. Two algorithms can reconstruct original information from vectors, which I accuracy, so securing them is as important as protecting the source data.

00:11:47:22 - 00:12:17:02 Unknown Got it, got it. But talking about security and challenges, what is the biggest mistake organizations make when connecting AI to their systems and their data? Not applying the least privilege principle. I walk flows and system should operate with the minimal privileges. This reduce the attack surface and prevents accidental access to sensitive data. Are there some industries more critical than others for AI security?

00:12:17:07 - 00:12:48:01 Unknown Well, all industries need security. But as cars, finance and government are particularly sensitive due to privacy regulations, even retail and logistic benefits from strong AI security practices. The risk level varies, but the principle remains you can't innovate safely without protecting your data. I agree, Thomas, but what if an AI system accidentally exposes sensitive data? Whose responsible and how should it be handled?

00:12:48:03 - 00:13:16:15 Unknown Zero risk doesn't exist. Security is always evolving. Auditing is key. Even in a breach, it tells you what happens from where and by whom. Combined with the access control, monitoring and training, auditing helps organization learn from incidents and adapt their defenses. Students don't need to have a learning environment to practice their skills. Oracle Academy provides such skilling opportunity.

00:13:16:15 - 00:13:49:22 Unknown Isn't it? What resource, course or lap do you recommend students to start experimenting safely? Definitely. I recommend visiting live labs. There are over 900 free workshops you can run in a sandbox on the Oracle Cloud. A safe way to get hands on experience with AI vectors and database security. So just search for AI vector or security reveals many interesting workshops to explore, and you may also get access to our environment through the Oracle Academy program.

00:13:49:22 - 00:14:10:16 Unknown But for this week, I think you will be able to guide them. Absolutely. I will do this with pleasure for you, us, and obviously also for the listener. I will provide some links at the end of this podcast. Thomas, what skills should students focus on if they want a career in AI security? Well, the core I.T skills are still essential.

00:14:10:16 - 00:14:39:14 Unknown AI is built on computers, network storage, data model and data types. All the fundamentals. Without these, you can't accurately judge what an AI system is producing until you understand the fundamentals. You can safely explore AI models, vectorization, and workflows without introducing security risk. And when it comes to security, missing the foundational concepts makes it impossible to build something solid.

00:14:39:16 - 00:15:13:06 Unknown Master the basics. There the foundation AI security stands on. So from there, student can expand their career and learning path in AI security. First, hands on experience use Sandbox on Brenneman like Oracle Live Labs to experiment safely. Build small AI workflow, vector search application or test access control practice experience. Teach more than theory alone. Second, certifications consider Oracle Database cybersecurity or AI ethics certification.

00:15:13:06 - 00:15:49:12 Unknown They demonstrate knowledge and open doors to specialized roles and also cross disciplinary learning. AI security sits at the intersection of database management, cybersecurity, and data science. So learn the basics in all three understanding vectors, embeddings, and database and its security layer domains. And we're not there yet. But thank you so much for sharing your expertise today. Interesting. It's been fascinating to see how AI and database security intersect, and your examples make these concepts very tangible domains.

00:15:49:14 - 00:16:16:15 Unknown Thank you. Rick. I really enjoyed the conversation. Security isn't just a constraint. It's an enabler for innovation. Designing systems with security in mind lets you safely leverage AI to unlock real value from data. Here are a few practical takeaways for our listeners. First, understand your data. Know what is sensitive, how it is classified, and what regulations apply.

00:16:16:15 - 00:16:49:02 Unknown Governance is the foundation of all security strategies. Second, master the fundamentals, networks, storage, data types and database principle form the essential basis for AI security. Then use security as a tool, not a barrier. Access controls, data driven security and auditing are enablers for confident innovation. Then experiment safely. Sandbox Live Labs anonymized data set lets student explore end zone without risk.

00:16:49:04 - 00:17:15:10 Unknown And last but not least, stay curious and proactive. AI insecurity evolve rapidly. Keep bringing in. Anticipate all new technologies can help or your systems. Absolutely. I would call that great advice, Thomas. And to our listeners, whether you are a student exploring AI or an educator or a professor instructor guiding the next generation. Security is definitely part of that.

00:17:15:10 - 00:17:46:11 Unknown Journey is not an obstacle. Exactly. Explore experiments safely and keep data protection at the center of your work. Secure data is usable data, so push the boundaries responsibly. And I can truly become a powerful tool in your toolkit. Absolutely. Thanks again Thomas. We hope to have you back soon for another discussion. So wrapping up today, we explored how AI is both a powerful opportunity and a security challenge.

00:17:46:11 - 00:18:13:06 Unknown Why the Oracle database is such a strong foundation for protecting sensitive data, and how features like access control, auditing, and vector security makes AI safer to use. We also touch on the role of governments, the risk of unrest, debating factors, and why mastering IT fundamentals is the best path into a career in AI security. So stay curious. Keep experimenting safely.

00:18:13:08 - 00:18:55:04 Unknown And remember, security isn't a barrier. It is what makes innovation possible. Curious and staying secure. A few links, like Thomas said on this topic please go to livelabs.oracle.com. And also for the maximum security architecture go to oracle.com/ To our listeners I encourage you to visit our website academy.oracle.com to learn more about Oracle Academy for educational institutes of all levels worldwide.

00:18:55:06 - 00:19:21:08 Unknown I also encourage you to subscribe to our podcast. Feel free to share this with your educational friends as well. Also, please follow Oracle Academy on Instagram and Facebook just to keep up with all our activities and with our members. Dear listeners and thank you, Thomas. Again, thank you so much. This wraps up this episode. Thanks for listening and stay tuned for the next Oracle Academy Tech Chat podcast.

00:19:21:10 - 00:19:33:12 Unknown That wraps up this episode. Thanks for listening and stay tuned for the next Oracle Academy Tech Chat podcast.