Your data is my data now

Most people see AI as a fancy search engine. You ask a question, it finds an answer.

That’s not how it works. It doesn’t retrieve information. It generates it.

To generate anything, it has to learn first. And what better way to learn than all the data everyone in your organization feeds into it?

Unless you specifically change the settings, AI systems like ChatGPT will use everything you give it as training data.

Once it has learned something, there’s no way to remove it. What it learned from you will be available to me, your closest competitor, and anyone else who asks the AI the right questions.

You might be thinking, “There are ways to give an AI limited access to information.” That’s true. There are two ways to do that:

The first option is fine-tuning a model. This is what customer service bots are based on. They’re primarily working off of the base model, but they have access to a library of unique information (like product details).

The second option is Retrieval Augmented Generation (RAG). This is how most enterprise AI tools work. Your data stays in your database, but the AI can query it when needed and then forget what it saw.

All of these ways AI can access your data come with security issues.

Closely guarded proprietary information has become standard training data because employees didn’t understand what happens when they ask ChatGPT to troubleshoot code.

Private information has been leaked when attackers asked a helpful chat bot a few well-crafted questions.

AI tools with access to entire databases have been used as convenient attack paths to otherwise well-secured organizations.

Traditional security measures fall short when it comes to managing AI security. Yet, you can’t blame other people’s ignorance or the tools’ design flaws when things go sideways.

Many organizations recognize that even experienced security professionals lack the expertise that's now needed. But even if you know what's needed, do you have the credibility that makes others listen to you?

That's the gap AAISM fills.

If you’d like to lead AI security, ISACA’s AAISM certification can give you a massive edge at the moment.

It’s the only AI-security management certification. Pass the exam, and you’re the most qualified person in almost any organization to manage AI security.

The exam is tough, but like with other certifications, nearly all our students pass on their first attempt. So, if you want to understand AI-security management, join our 3-day Bootcamp.

Enroll today for the May Bootcamp and save $300.

Stay secure,
The DestCert Team