IBM

Since the rise of generative AI, there's been a growing concern about the proliferation of large language models and what that means for security and compliance teams. As one of its New Product Introductions, IBM set out to build an AI Security product as part of its Guardium data security portfolio.

The company needed to understand what the new era of generative AI looked like in the world of security - before attempting to build anything. I set out to get answers.

Goal:

• Understand how security teams are securing AI
• Understand how governance and compliance are affecting AI security
• Gauge the viability of IBM's proposed AI security offering

Method: User interviews.

Participants: Eight customers, with job responsibilities that involve securing generative AI models.

The work to understand the AI security landscape and the product offering was broad. We gleaned many insights that are classified around themes of: Complimentary Strategy, Security Environment, Regulatory Environment, and Product Capabilities.

The complimentary strategy proposed by IBM, would enable current IBM customers who use Guardium AI Security and Watsonx.governance, to maximize the offerings and achieve a robust AI security and AI governance solution. This strategy was validated, as participants indicated that the synergy achieved would provide maximum value.

These insights are classified under the theme of Security Environment

Threat management is a struggle

All expressed difficulties in detecting and mitigating threats related to AI technologies.

Nearly all mentioned how susceptible AI technologies are to cyber attacks.

Recommendations

Considering that customers will be starting from a blank slate, they will begin with a voluminous amount of threats detected. Providing the level of criticality will be table stakes in order to reduce alert fatigue and offer basic direction.

Providing additional ways to prioritize what to remediate will drive real value.

"It's like we're practically defenseless with this enemy."

Proactive security is in its infancy

Some participants have undergone AI security training.

Some have implemented a form of continuous monitoring of AI models.

Tools mentioned include: Wiz, Datadog, and Dynatrace.

Recommendation

The situation leaves much room for adoption and growth, with the advent of an advanced AI security tool.

It also reveals that competition is already heating up, making time-to-market of paramount concern.

"It's gonna take us a while to get there, but we're trying."

Prompt injections drive fear

Most participants specifically mentioned prompt injections as a major concern that they are not prepared for.

Recommendation

Any AI security tool developed will need to meet this critical user need.

"It's gonna take us a while to get there, but we're trying."

RBAC used as a backstop

The majority of participants rely on RBAC as a way to mitigate risks associated with accessing AI technologies.

Recommendation

Although RBAC is important to any security strategy, customers will require more advanced methods to effectively secure their AI environments, leaving Guardium AI Security with an exceptional opportunity.

Consider an awareness campaign to educate customers on what is required to secure AI, along with the risks of relying on RBAC.

"RBAC isn't gonna solve all our problems, but it's an essential part of the foundation."

Data protection is at the core

All participants emphasized relying on encryption and anonymization to secure sensitive information and training data.

Recommendation

This positions IBM Guardium very well, being a native data security platform with a forthcoming quantum-safe encryption tool.

Tight integration of Quantum Safe Posture Management into AI Security can drive upsell and cross-sell of the product.

"We rely on strong encryption to help us with this problem."

Regulatory Environment

These insights are classified under the theme of Regulatory Environment.

AI governance takes center stage

Most participants mentioned AI-specific governance frameworks, such as the EU AI Act and NIST, AI RMF which will bring a paradigm shift in how they secure AI and demonstrate compliance.

Recommendation

Providing educational resources and a guided compliance journey similar to what is offered in Guardium SaaS, will provide both ease of use and increased velocity when developing and executing an AI compliance program.

"The GRC team is talking to us about the EU AI Act. That's great - but we've got to figure out the details on the ground."

Collaboration on compliance isn't optional

The majority of participants reported regular engagement with legal and compliance teams concerning AI.

This underscores the necessity of cross-functional collaboration in navigating the complex regulatory environment surrounding AI.

Recommendation

Providing a compliance workspace similar to what is offered in Guardium SaaS, will allow for ease of reporting and cross-functional collaboration on AI compliance.

"This is becoming more and more of a team sport."

Global compliance complexity persists

Most participants expressed challenges associated with managing global compliance.

An effective AI security tool must accommodate the nuances of operating across different global regulatory jurisdictions.

Recommendations

Provide support for a diverse set of global AI regulations as they become available.

Develop a regulartory roadmap that serves as a blueprint for when regional and global compliance mandates will be added to the product.

"This is all moving at different speeds across the globe."

Product Capabilities

These insights are classified under the theme of Product Capabilities.

Complexity demands simplification

Several participants expressed a desire for simplified tools that do not carry a high learning curve.

Recommendations

Leverage existing, battle-tested design patterns found in other Guardium products.

Explore ways to increase automation in the product.

"We're about to have AI throughout the entire environment. We need simple and streamlined tools if we're ever gonna have a fighting chance to actually be successful."

Content can drive understanding

Most participants desire technically accurate, yet easy-to-understand labels and terms.

There is a concern that an advanced tool in the uncharted world of AI could arrive with an excessive cognitive load attached.

Recommendations

Put an emphasis on content design, with dedicated staff supporting the AI security product.

Ensure that content is a mandatory part of usability testing.

"We've got a lot of different teams working on this. And everyone doesn't operate at the same technical level. Just give it to us straight."

A single pane of glass is needed

All participants expressed a desire to employ a single solution for AI security.

There is a desire for a tool with comprehensive visibility into all AI models in use, including the capability to detect Shadow AI projects.

Recommendations

Tight integration with Guardium's Data Security Posture Management (DSPM) solution.

Consider adding on-premises functionality.

"I need a solution that does it all. I can't be adding another ten tools just for AI."

Visibility isn't enough

Most participants desire a tool that goes beyond visibility and aids in remediation.

Recommendations

Telling users exactly what needs to be done to remediate is table stakes.

Adding automation to speed up remediation would drive an incredible amount of value.

"This is nice, but I need more. I need help actually fixing things."

With our new understanding of the landscape, along with the user mental model of AI security practitioners and their unique needs, we built an enterprise-grade solution that allows security and governance teams to work together to secure all of their AI.

AI Security Dashboard

The AI Security Dashboard provides full visibility into the AI environment. Whether it's generative AI, shadow AI, or agentic AI - discover it all and manage it under a single pane of glass.

AI SPM

The AI Security Posture Management tool discovers vulnerabilities and misconfigurations in AI - providing insight into what to remediate next.

AI Gateway

The AI Gateway tool enables the creation of global policies for prompt protection.

Integrations

We created integrations that provide synergies for security and governance teams - from Watsonx.governance to OpenPages - security and GRC teams are finally unified.

The AI Security endeavor was an eye-opening experience. While generative AI is indeed a new and fast-moving space, the intense desire amongst participants to secure AI with a comprehensive solution was even more than expected.

It was this desire, along with discovering their unique needs, that drove the team to create something great for our customers. It's a prime example of how empathy can drive innovation, even in unchartered territory.

Next will come usability testing. There we will gauge if our customers can effectively use the novel product that we developed.