Download our new Personal Protection from CyberCrime ebook for free!

Guest Blog
by
Sean Todd

AI in Security

Letting Humans Focus on What Matters

In today’s fast-paced digital environment, information security teams are inundated with large amounts of data. Maria's team struggles daily to keep up with thousands of alerts and logs, highlighting the overwhelming challenge faced by many professionals in the industry. The challenge is exacerbated by the growing complexity of threats, forcing security professionals to sift through vast amounts of logs, alerts, and other data to detect vulnerabilities. This tedious work often leads to missed details and burnout, weakening overall security postures. Enter artificial intelligence (AI)—a technology poised to revolutionize how security teams manage and prioritize threats. This article explores how AI-backed tools can transform information security processes, allowing professionals to focus on higher-value tasks and expanding organizational defenses.

AI and Threat Detection

Information security teams are tasked with monitoring all possible attack vectors at their organizations which means that they must sift through the modern firehose of log data generated every second of every day. Of course this is an impossible task with limited resources. Skilled security professionals will still feel the pressure to do a perfect job because any misses could lead to a company-killing data breach. This monumental task, coupled with the pressure to achieve perfect results, often leads to lower morale and burnout. 

AI tools present a unique opportunity to tackle these challenges. By automating tedious and repetitive tasks, AI allows security teams to redirect their efforts towards more impactful activities, such as advanced threat hunting and strategic planning. Human analysts may take days to manually review documents or logs, needing regular breaks for rest. AI-backed tools though can quickly and accurately sift through large datasets, detecting patterns that indicate potential threats, all without needing to ever take a break. Humans struggle to maintain contextual awareness across all organizational assets, often resulting in gaps in security coverage. AI excels at maintaining a holistic view, providing consistent vigilance without these human limitations. They can cross-reference threat intelligence feeds with internal log data and understand how all the pieces of the infrastructure interact. These advantages can help in identifying tactics, techniques, and procedures (TTPs) commonly used by adversaries such as advanced persistent threats (APTs) or state-sponsored attackers.

These tools not only speed up analysis but also make connections across disparate datasets more effectively than humans. They promise to identify large-scale attack patterns that might otherwise be overlooked. Importantly, AI eliminates inherent biases that may cloud human analysis, allowing data to be viewed without the preconceptions that can lead to blind spots—such as overlooking an internal threat from a trusted colleague.

Shifting Roles in Security

As AI assumes more of the repetitive work, the role of security professionals is bound to shift. Senior specialists will be free to focus on proactive defense strategies and security architecture. For entry-level roles, this transformation could mean a pivot from manual data review to developing, fine-tuning, and instructing AI models. Security work in general will move beyond reacting to constantly evolving security threats and into a more cerebral field focused on overall strategy.

This shift also brings new opportunities. Early-career professionals will learn how to operate in an AI-driven environment, gaining experience in leveraging AI tools alongside traditional security strategies. In the long run, this will help incubate a new generation of security professionals adept at managing threats and utilizing AI to augment their capabilities. More novice security professionals will have the opportunity to look beyond the traditional entry-level tasks of log watching and tool configurations.

Trust, Transparency, and Testing

AI tools are not without risks and not without warranted mistrust. False positives and overreliance on AI-generated insights can create significant issues, especially when teams blindly trust these systems. Therefore, it is crucial for security professionals to remain actively engaged in evaluating AI decisions, ensuring accuracy and reliability. A false positive, for instance, can lead to unnecessary disruption, wasted resources, or even loss of trust in the AI itself. To avoid this, human oversight remains critical to evaluate and verify AI outputs. Humans must remember that AI tools are inherently non-deterministic. While this enables AI to creatively address problems, it can also lead to unreliability if not properly managed.

To maintain trust in AI systems, organizations must hold their vendors accountable. For each of your vendors using AI, you should

  • Insist on transparency, ensuring that the data fed into AI models and any additional logic are accessible for inspection.
  • Demand clear documentation on AI decision-making processes to understand its behavior and outcomes.
  • Prioritize explainability to understand why and how AI reaches conclusions.
  • Ensure that your company's data is not fed back into the models where other clients may be able to extract it.

Regular testing of AI systems is essential, using known datasets to ensure accuracy. Feed your AI tools data with expected outputs and make sure that it returns a reliable answer. A red team could accomplish this by playing the role of adversary in your network and see if the blue team's tools can still see them or if it miscategorizes the actions of the other team.

Adapting to  AI-Driven Decision Making

AI systems are quickly evolving towards a future where they can not only pattern match and just regurgitate data, they will soon be capable of using logic and reasoning. The recent release of the o1 model from OpenAI shows just how far LLMs have grown in the last few years. These new models will be able to further enhance a security team's ability to extrapolate novel exploits in company systems. Security leaders will be able to quickly generate complex but plausible threat scenarios that can not only be used to harden systems but to train others in how bad actors see the company's systems.

Right now, incident response training usually involves a tabletop exercise derived by one of the security team members. These models hold the promise to move beyond that. Feed into it data about company systems and personnel and the models will be able to create diverse scenarios such as internal threats, misconfigurations, or ransomware. Because of the time savings, security professionals will also be able to create more frequent and more tailored exercises. Rather than a single overarching scenario annually, the security team will be able to add in smaller trainings focused either on a specific product feature or a certain high risk team.

Quantifying AI's Impact

With AI-backed tools aiding security teams, how can we measure their effectiveness? After all, your company's finance team will want to know that they are getting their money's worth from these tools. While measuring security programs has always been a challenge, it doesn't have to be the same for these new tools. Look at the current state of your security program and estimate the percentage of company assets and processes that are not closely monitored by the security team. I would wager a guess that there are a lot of lower priority assets that are not monitored closely. Now measure how many of those forgotten systems your team can watch given these new AI-backed tools.

Conclusion

AI holds immense promise for information security teams, helping to reduce the burden of manual data review and freeing professionals to concentrate on strategic, high-value tasks. By automating tedious work, AI allows security teams to shift their focus toward proactive defense strategies, ultimately bolstering their organization’s security posture. However, achieving these benefits requires careful integration, as well as ongoing training and testing to ensure the AI tools work effectively and are not misused or over-relied upon.

The shift in roles that AI brings will present exciting new opportunities for security professionals. Junior analysts will have the chance to develop skills in managing and instructing AI systems rather than being confined to repetitive data review tasks. This transformation can cultivate a new breed of InfoSec professionals who are highly skilled in both traditional security techniques and the deployment of cutting-edge AI technology.

Ultimately, AI-driven tools have the potential to reshape the future of information security by extending the capabilities of security teams and enabling more comprehensive monitoring. By thoughtfully adopting and integrating AI, organizations can significantly enhance their defenses, reduce the likelihood of burnout among security professionals, and build a more resilient security posture for the future.

About our Author
Sean Todd

Sean has been practicing information security for 12 years and in that time has built up the security programs from scratch for three companies. In addition to his primary work duties, he consults on the side for early-stage security startups. Currently he works as CISO and lead AI dev for Auditive.io, a first of its kind two-sided TPRM platform with continuous monitoring.

Connect with
Sean Todd
on
LinkedIn

Still have questions?