Traditional AI and machine learning have been around for quite some time, helping to develop models, analyze data, make predictions, and automate tasks. Generative AI has now taken the technology world by storm with its rapidly scaling ability to create everything from essays to videos to software code based on massive sets of training data.
The global AI market is projected to pass $150 billion in 2023 and grow at a CAGR of 36.8% from 2023 to 2030, according to MarketsandMarkets Research. Some of this technology’s power has become benignly commonplace, like unlocking our phones with facial recognition, but a new breed of AI that can create content is more of a double-edged sword.
AI allows businesses to transform themselves into more efficient entities both in the way they operate and in their customer offerings. Not surprisingly, cybercriminals are doing the same thing, using AI to become more efficient at attacking businesses and governments.
The U.S. Government recognizes the magnitude of the threat, illustrated by its launch of an AI cyber challenge with a $20 million prize purse. And, according to GlobalData, more than 30% of communication service providers (CSPs) have experienced eight or more breaches in the last year. So, with the global average cost of a data breach hitting $4.45 million this year, everyone wants to find ways to be more secure and stay a step ahead of cybercriminals who are increasingly using AI to intensify their efforts in a few key ways.
The Role of AI in Cybercrime Today
The first technique revolves around spear phishing, a highly targeted type of phishing. AI scans hundreds of thousands of social media profiles to identify suitable targets. Then generative AI kicks in to personalize interactions based on what it learned from those profiles. Those can be emails that impersonate the CEO of a company or deepfake videos that appear to be a company executive, luring victims into granting access to systems and information.
A second method is using AI to disguise activities. Machine learning algorithms can be manipulated to “poison” AI training data sets, causing AI-based security systems to classify potentially malicious activities as benign. AI-powered malware can study defenses and imitate everyday system communications until the right opportunity presents itself. For example, rather than tricking people into giving up passwords, malware code can be programmed only to activate when facial recognition is used to gain access to a system or specific data.
The third category involves harnessing AI to create malicious code or to reverse engineer software code to infiltrate systems, like networks. Once code is written and tested, generative AI can translate it into a variety of programming languages to attack other systems. With reverse engineering software to find vulnerabilities, generative AI can speed up the process, so attackers can get a high-level view of code functionality and determine an area to target more quickly.
The Power of AI in Cybersecurity
Fortunately, AI in cybersecurity has also risen in prominence and is allowing businesses to analyze large volumes of data in real time, identifying subtle patterns and anomalies often missed by manual detection to improve accuracy.
Speed is just as crucial for attack responses. Businesses leveraging AI-powered security solutions have been able to identify breaches more than 100 days sooner and preserve up to $1.76 million compared to those that do not harness AI technology, according to IBM. In addition to improving security, AI can reduce costs related to workload and resources.
Generative AI can take things a step further, overcoming complexity related to methodology and tools used to protect against cyberattacks. It eliminates the need to learn complicated query languages and complex operations, turning even entry-level security analysts into security superheroes.
Improving Security Software Creation With AI
Including AI in the software creation process makes it possible to create and optimize test cases on a wide range of scenarios, allowing for better improvisation during an attack. Generative AI can augment that capability, producing large and varied training data for the test cases.
A hallmark of AI is its ability to make predictions based on learning. In the case of security, this means analyzing past breaches to predict where subsequent attacks will occur and what trajectory they may follow. To assess the resiliency of the software and the processes in place for security analysts, generative AI can play the role of the hacker, creating attack simulations that expose more potential areas of vulnerability.
That learning capability also means continuous software improvement, building upon knowledge over time, and creating faster feedback loops to eliminate vulnerabilities. Using historical software data, user feedback, and industry knowledge as training data for an AI model, developers can employ AI solutions to assess design ideas, assist in prototyping, and automate error-prone repetitive tasks.
While developers still need to check the code, harnessing generative AI as part of their development process ensures better software quality, building trust with customers.
About the Author
Raghav Sahgal is a global leader and entrepreneur in the cloud, software, and communications space, paving the path to the future where networks meet the cloud.
In his current role as President of the Cloud and Network Services business at Nokia, Raghav leads the transition for CSPs and enterprises across the world to 5G, secure autonomous operations, and industrial digitalization. He is passionate about helping them embrace new business models and value-creation opportunities with digital ecosystems.
As former President of Nokia Enterprise, Raghav positioned the business group as one of the market leaders in private wireless networks and achieved significant growth, delivering mission-critical networks and digital automation solutions to enterprises, including webscale and industrial businesses, as well as government entities.
