What Role Is AI Playing in Cybersecurity in 2023?

The development of AI has been a game-changer for nearly everyone, and that fact is no different in the world of cybersecurity. New threats powered by AI are reshaping traditional attack vectors, including cryptography, prevention, and social engineering.

In this article, we’re discussing how, in the so-called AI Boom of 2023, cybersecurity is being shaped by this new technology.

Artificial intelligence (AI) is a rapidly evolving technology that significantly impacts many industries, including cybersecurity. AI can automate tasks, identify and respond to threats, and improve decision-making.

In cybersecurity, AI is used in a variety of ways, including:

• Malware Detection: AI analyzes code or behavior to spot malware. Machine learning algorithms learn from known malware patterns and can even catch new, previously unseen malware.
• Phishing Spotting: AI scans the content and structure of emails to flag phishing attempts. It uses natural language processing to pick out suspicious keywords and phrases. It can also spot phishing emails coming from hacked accounts.
• Intrusion Prevention: AI monitors network traffic to catch unauthorized system access. It can also detect threats originating from within an organization.
• Risk Evaluation: AI assesses the likelihood of a cyberattack by studying past attacks and the current state of an organization’s security. It can also pinpoint potential weak spots that attackers might exploit.
• Incident Handling: AI automates the response to cyberattacks. This includes sorting incidents, deploying fixes, and recovering from damage. It can also forecast the impact of an attack and prioritize actions accordingly.

Even though AI is a relative newcomer in cybersecurity, it’s already making waves by transforming how we secure our systems and data. As AI tech advances, expect it to take on an even bigger role in cybersecurity.

Why Use AI in Cybersecurity?

The unique blend of autonomous action and advanced pattern recognition makes AI suitable for several cybersecurity applications. These include:

• Scalability: With AI, you can scale up to manage huge data volumes and increasingly complex threats, a crucial ability as cyberattacks evolve.
• Speed: AI spots and counters threats much faster than a human could, a vital feature in a world where cyber threats can strike any moment.
• Precision: AI outperforms humans in threat identification because it can learn from a far larger data set.
• Cost Savings: AI automates tasks that would otherwise require human intervention, making it a cost-effective solution for enhancing cybersecurity.

Challenges of Using AI in Cybersecurity

On the other hand, using AI in cybersecurity has its challenges. Technology is flawed, much less AI, which can cause issues when adopting this approach. 

Some challenges include:

• Complexity: Understanding AI systems can be tricky, making troubleshooting and ensuring proper function challenging.
  
• Bias: AI can produce false positives or negatives due to built-in biases, which either unnecessarily shut down a system or let a cyberattack slip through.
• Data Needs: Training and operating AI systems require a lot of data, posing a challenge for organizations with limited historical data.
• Regulatory Challenges: Regulatory requirements can slow down AI adoption as authorities work to ensure the technology’s safe and responsible use.

Despite these obstacles, AI holds great promise for significantly improving cybersecurity. As the technology matures, its impact on cybersecurity will become even more substantial.

Artificial Intelligence in the Hands of Malicious Actors

AI isn’t just a tool for defenders. Hackers and other malicious actors are delving into AI and machine learning to power all sorts of new attack techniques and scams. The sheer breadth of how AI may be applied to cyber threats may reshape the cybersecurity landscape. 

Some of these attacks include:

• Stealth: Using AI, hackers can remain dormant within a network for extended periods, setting up backdoors to critical infrastructure. This allows them to launch more sophisticated attacks when they are ready.
• Deepfakes and Impersonation: AI enables hackers to create deepfake data and fake human interactions on social networking platforms. This can be used for social engineering attacks to manipulate individuals into divulging confidential information.
• Advanced Malware: AI helps build better malware that can adapt and evolve to bypass security measures.
• Password Cracking: Advanced password-guessing techniques and even CAPTCHA-cracking become possible with AI-supported algorithms.
• Weaponized AI: Hackers can turn AI frameworks into tools for breaking into vulnerable systems, opening a new front in cyber warfare.

What’s the Deal with Generative Adversarial Networks?

A Generative Adversarial Network (GAN) consists of two machine learning models: a generator and a discriminator. The generator crafts new data while the discriminator judges if that data is real or fake.

Both the generator and discriminator train together. The generator aims to produce data so convincing that the discriminator can’t tell it’s fake. Meanwhile, the discriminator works to separate real from fake data. As they train, both get better at their jobs. The generator becomes more skilled at creating lifelike data, and the discriminator learns to tell real from fake.

GANs are often used in basic but complex processes that bleed into other industries:

• Image Creation: GANs can craft realistic images for various uses, like virtual environments, new product designs, or image quality improvement.
• Text Crafting: From poems and code to scripts and music, GANs can generate text for creative content or realistic chatbot dialogue.
• Voice Production: GANs can produce lifelike human voices for audiobooks, movie dubbing, or realistic virtual assistant speech.

More specifically, in the area of cybersecurity, GANs have had an impact on both sides of the line (attacker and defender).

Defensive Applications of GANs in Cybersecurity

• Anomaly Detection: Traditional anomaly detection systems can be limited by the quality and quantity of the data they are trained on. GANs can generate synthetic data that mimics normal network behavior, providing a richer dataset for training more effective anomaly detection systems.
  
• Data Augmentation for Threat Intelligence: GANs can augment existing datasets with synthetic yet realistic examples of cyber threats, thereby improving the performance of machine learning models used in threat detection.
  
• Phishing Detection: GANs can be trained to generate phishing attempts that are then used to train anti-phishing algorithms, making them more robust against sophisticated phishing attacks.
  
• Simulating Cyber Attacks for Training: GANs can simulate realistic cyberattacks, providing a safe environment for cybersecurity professionals to train and prepare for various attack vectors.

Offensive Applications of GANs in Cybersecurity

• Advanced Malware: GANs can create malware capable of adapting and evolving to bypass security measures, making it more difficult for traditional antivirus programs to detect and remove them.
  
• Data Poisoning: GANs can generate synthetic data that looks legitimate but is designed to mislead or corrupt machine learning models, thereby affecting their performance and reliability.
  
• Deepfakes: GAN-generated deepfakes can be extremely convincing, making it easier for attackers to impersonate trusted individuals and deceive targets into revealing sensitive information or performing actions compromising security.
  
• Evasion Techniques: GANs can be trained to understand how intrusion detection systems work and generate network traffic patterns that are difficult to distinguish from legitimate traffic, thereby evading detection.
  
• Automated Vulnerability Discovery: GANs can automatically discover new vulnerabilities in a system, providing attackers with more opportunities to exploit.

What Does the Future Hold for Cybersecurity?

With AI disrupting industries across the board, we’re not surprised to see cybersecurity impacted as well. So what do you do?

The truth is that a good defense is still possible, particularly for those who support security and compliance best practices. No protection system is foolproof, but with the right partners, you can make sure your infrastructure is ready for any challenge. 

That partner is Lazarus Alliance.

What Is Autonomous Malware?

We’re reaching the end of 2025, and looking ahead to 2026, most experts are discussing the latest threats that will shape the year ahead. This year, we’re seeing a new, but not unexpected, shift to autonomous threats driven by state-sponsored actors and AI.  With that in mind, a new generation of threats, broadly known as...Continue reading→

What CISA’s Emergency Directive 26-01 Means for Everyone

In mid-October 2025, the CISA issued one of its most urgent orders yet: Emergency Directive 26-01. The directive calls on all Federal Civilian Executive Branch (FCEB) agencies to immediately mitigate vulnerabilities in devices from F5 Networks following a state-sponsored breach of F5’s systems and access to portions of BIG-IP source code and vulnerability data. The event...Continue reading→

Cybersecurity and Vetting AI-Powered Tools

A recent exploit involving a new AI-focused browser shone a light on a critical problem–namely, that browser security is a constant issue, and AI is just making that threat more pronounced. Attackers discovered a way to use that browser’s memory features to implant hidden instructions inside an AI assistant. Once stored, those instructions triggered unwanted...Continue reading→

Shutdown Security And Cyber Vulnerability

When the federal government shuts down, the public sees closed monuments, unpaid workers, and halted programs. What they do not see is the silent surge of cyberattacks targeting agencies already operating on fumes. During the most recent shutdown, attacks against U.S. government systems spiked by nearly 85%.  Cybersecurity failures during government disruptions rarely start with...Continue reading→

Identity and the Shift from Malware

The world of cyber threats is rapidly evolving, and while we can see these changes more generally, it’s always crucial to understand them concretely. As the 2025 CrowdStrike Global Threat Report shows us, the landscape of our industry is changing.  We’re digging into this report to discuss a challenging trend: the move of hackers foregoing...Continue reading→

Maintaining Compliance Against Prompt Injection Attacks

The increasing adoption of AI by businesses introduces security risks that current cybersecurity frameworks are not prepared to address. A particularly complex emerging threat is prompt injection attacks. These attacks manipulate the integrity of large language models and other AI systems, potentially compromising security protocols and legal compliance. Organizations adopting AI must have a plan...Continue reading→

Are We Already Talking About CMMC 3.0?

The ink has barely dried on the CMMC final rule, and already the defense contracting community is buzzing with speculation about what comes next. Just when contractors thought they had a moment to catch their breath after years of regulatory limbo, whispers of CMMC 3.0 have begun circulating through the industry. But is this just...Continue reading→

Centralizing Identity-Based Risk

As the traditional network boundary dissolves and remote work becomes standard practice, identities are the major frontier for security. Whether we’re talking about human users, service accounts, or machine identities, these have emerged as both the primary access mechanism and the most targeted attack vector.  It has become imperative for providers to centralize identity management...Continue reading→

Deviation and Significant Change Requests in FedRAMP: A Comprehensive Guide

FedRAMP provides a standardized approach to security assessment, authorization, and continuous monitoring for cloud products and services used by federal agencies. While the program’s rigorous baseline requirements ensure consistent security, the reality is that this consistency calls for a little flexibility.  This is where deviation requests and significant change requests come into play. These two...Continue reading→

The Costs of Compliance and Data Breaches

Data is possibly one of the most valuable assets any organization holds. Customer information, employee records, and proprietary business intelligence present challenges because the data flowing through modern enterprises represents both significant opportunities and serious risks.  Businesses face a challenging balance: investing in compliance measures to protect sensitive information while also preparing for the real...Continue reading→