Is AI Friend or Foe to Cybersecurity?

The buzz surrounding artificial intelligence is echoing through most industries. Unlike some advancements that light up media channels only to quickly disappear, AI’s rapid development and adoption is changing tools and tasks.

AI isn’t new, but recent mainstream advances now put its usefulness at the fingertips of everyday users. More cybersecurity tools are incorporating it, and more cyber attackers are leveraging it. Knowing your risks and rewards on both sides of the AI coin is important.

There are currently seven main types of artificial intelligence known today:

1. Based on capabilities:
   1. Artificial Narrow Intelligence (ANI)
   2. Artificial General Intelligence (AGI)
   3. Artificial Superintelligence (ASI)
      
      
2. Based on functionalities:
   1. Reactive Machines
   2. Limited Memory
   3. Theory of Mind
   4. Self-Aware

Defining AI by capabilities

Narrow AI (Weak AI) is widely used to perform narrow tasks, such as facial recognition or web search, and powers automation that eliminates human intervention for daily or routine operations. Streaming services that suggest shows you might like based on viewing behaviors, shopping sites that suggest items based on previous purchases, and automated content and grammar tools are examples of Narrow AI.

General AI (Strong AI) will have human-like cognitive abilities, including common sense and reasoning and the ability to adapt and alter as circumstances change. This version is in the early development stages and has a significant distance to go before achieving the competency indicated in its definition.

Superintelligent AI is currently only theoretical and would herald the time when machines surpass human intelligence. This AI version is used in science fiction storylines regularly.

Defining AI by functionalities

Reactive Machines are AI systems that don’t store “memories” or past experiences to predicate future actions. This AI version analyzes and responds to different situations as they occur. One example is IBM’s Deep Blue, which famously beat Gary Kasparov at chess in the 90s.

Limited Memory types of AI can maintain records of past information and inputs. This category includes Generative AI tools commonly used in information technology (IT) and information security (IS), such as ChatGPT, Microsoft Copilot, Google Gemini, and IBM watsonx.

Theory of Mind is a more advanced type of AI, which is still a work in progress. This type of AI would be able to understand and remember emotions, beliefs, needs, and use that information to make decisions.

Self-Aware AI represents the future of AI. This type of AI is where machines have consciousness, sentience, and self-awareness. This technology is only theoretical for now but is a well-recognized theme in dystopian, end-of-the-world movies where robots take over the world.

These definitions highlight the two sides to everything—the good and the bad. Our ability to harness the good and prevent the bad is at the core of AI in cybersecurity.Two sides of AI for cybersecurity

At the start of 2024, our VP of Cybersecurity shared prognostications on how the cybersecurity landscape will look as the year unfolds. These forecasts included AI in the “unfavorable conditions” and “clearer skies” sections.

Why both? AI performs to the intent of the people building and using the tools, making it valuable to cyber attackers and cyber defenders. Now that we’re at the midpoint of the year, how accurate was our look ahead, and what’s changing?

AI weaponizes deepfakes

Deepfake video technology has advanced to the point that national media and governmental agencies are expressing serious concerns about its use and impact. Politics and public figures have been the primary targets, but the money-making side of the equation is growing. Not long after publishing our 2024 outlook, a Hong Kong employee of a multinational firm transferred $25 million as requested by the company’s CFO on a video call that included other company employees. The only real person on the call was the victim.

Deepfake videos are used to threaten, cajole, and extort money from businesses and individuals. Attackers use intricate operations like the one above and inexpensive AI tools to create convincing social media posts, emails, and audio recordings.

AI arms threat detection

Earlier this year, enhancing security technologies with AI to improve threat detection and response gained steam as more vendors incorporated it. In our expert’s opinion, the biggest winners in this category looked to be security orchestration, automation, and response (SOAR) technologies. The biggest challenge was centralizing, operationalizing, and capitalizing on the disparate security tools using AI.

Fighting fire with fire pits AI against AI. With AI, zero trust gets a boost by automating and orchestrating core tenets of identity and access management (IAM), threat detection, and security information and event management (SIEM).

Working with AI’s dual nature

In securing your organization against AI’s negative impacts, make sure that your security awareness training is current and includes specific information that educates your employees on how to spot deepfakes.

Harnessing the positives of AI is easier than ever, with ubiquitous use driven by large technology companies introducing and supporting their own AI models. These AI models perform different tasks for specific use cases. Some of the biggest players are:

Microsoft is a front-runner in AI, with billions spent partnering with OpenAI to improve Bing’s search engine capabilities. Microsoft has also added AI functionality to its Edge browser and its Defender and Purview suites via Copilot for Security. Converge is a Microsoft Solutions Partner and has Copilot subject matter experts.

Other security partners integrating AI into their products include CrowdStrike with Charlotte, Sumo Logic dashboards, and Palo Alto Networks Precision AI. Another AI technology forerunner is IBM with Watson, now iterated with Watsonx for enterprise AI builders. Watsonx for Governance helps mitigate risks, perform lifecycle governance, and streamline compliance. Converge is an IBM Platinum Partner.

Converge’s AI partner ecosystem

The list of tech companies investing in the AI space goes on well beyond that list. The depth of our partner portfolio means we have established relationships with most of them, providing us with up-to-date training and information on the latest AI advancements.

Converge’s partner ecosystem puts us on the front lines of understanding, incorporating, and protecting our partners’ AI tools and your environments, data, and users. We invest in the burgeoning AI space across core areas, with our AI for cybersecurity expertise empowering us to guide, implement, and secure AI in client environments.

AI is here to stay, and Converge is here to help you deliver on its promise.