Artificial Intelligence in Cybersecurity: Fighting Cyberthreats in the Modern World

Artificial Intelligence in Cybersecurity: Fighting Cyberthreats in the Modern World

AI has existed since the mid 1950’s, but only recently has its popularity grown as it has in the last fifteen years. One example of this is its exponential use in cybersecurity. Anomaly detection, eradication, and predictive analytics are now integral parts of many AI-powered security systems. More advanced cyberattacks call for more advanced defenses, and AI is the next step up in cybersecurity.

With the rise of AI, “Cyberattacks are no longer solely the domain of skilled and specialized individuals.” (Ghana) This means many more individuals than before can successfully pull off an attack against a company, and they no longer must operate in groups. Additionally, “With the remote work continuing throughout several organizations even after the pandemic, the attack surface has widely expanded.” (Ghana) With AI, attackers can create very believable phishing attacks that can fool even security-aware employees. AI-enhanced malware can operate independently and adjust its behavior to better trick the system, and vulnerability scanning has been automated and optimized with artificial intelligence. Ghana Times warns that “[AI] can disable security alerts, mimic normal traffic, and re-establish footholds if eradicated.” (Ghana) AI also assists criminals in transporting pilfered data without interception. According to Ghana Times, “In ransomware, AI selects the most valuable assets to encrypt, ensuring maximum pressure on victims.” (Ghana) The more valuable the data is, the more likely the company is willing to pay to get it back. A survey conducted by Sophos says that “49% of victims paid the ransom to get their data back... it is the second highest ransom payment rate in six years,” Sophos) while the amount paid has dropped by half. Using AI, these attacks take less time to complete, allowing the attacker to pull off multiple attacks against different targets in hours instead of weeks, meaning victims must fight many more attacks in a shorter amount of time. Hackers use AI to scrape “public sources, social media, and organizational websites to gather intelligence on targets,” (Ghana) enabling attackers to find information much more quickly.

When defending against AI powered threats, companies must “fight fire with fire.” AI can aid an organization in many aspects of defense. Artificial intelligence can automate numerous systems such as malware and phishing detection and containing a successful breach. It can also automatically send out alerts when it believes a threat has occurred. Defensive AI uses many of the same techniques as hackers do. It can scan for vulnerabilities and report them and will scour the internet and “analyze dark web chatter looking for new exploits,” meaning it will look at what others are saying or doing to find the cracks in their own defenses.  (Monitor) Another capability of AI is its ability to “create highly realistic simulations of cyberattacks,” (Fortinet) enabling companies to find weaknesses in their defenses. An advantage of using AI to automate security systems is its ability to gather and analyze vast amounts of information in a relatively small amount of time. It can gather info on attacker behaviors and react to matching patterns by analyzing the behaviors of employees, the AI can spot anomalies and investigate them and identify “weak links such as exposed employee information or vulnerable applications” (Ghana) such as “unusual login attempts, and abnormal traffic”. (Fortinet) Another advantage of artificial intelligence is its ability to monitor systems 24/7 and never get tired. Companies can use AI to scrape the web and dark web for information on themselves and fix any vulnerabilities they might find before hackers or cybercriminals get to it.  

The military also uses AI in their security systems. The AI can be fed gathered intel on other countries and attack patterns that are not available to the public, making it more effective and accurate. This kind of AI has access to classified intel, making it easier for it to defend against foreign threats and sophisticated attacks by another country. It is necessary for governments and the military to have this level of security, as they are usually the ones with the most valuable or classified information and the most likely to be attacked by another country. AI is used by the military to create a realistic simulation of a cyberattack, to find vulnerabilities, and prepare for a real-world attack. A fast response time is critical in a military operation, and artificial intelligence can cut the time by 60% compared to traditional systems, especially hybrid AIs, which are a combination of multiple types of AI. Hybrid AIs are generally more adaptable and flexible but are also more complex and less transparent. Using HITL (Human in the Loop) inserts an employee into the system to allow more monitoring and supervision. It also can create more trust between the customer and company if there is someone closely monitoring the AI, especially important tasks and direct contact between the user and the AI, such as customer support chatbots.

AI is not perfect, so there are limitations to its capabilities, as Jenna Kerwin states, “Cybercriminals develop new and dangerous methods of cyberattack every day.” (Kerwin) AI sometimes cannot keep up with new malware strains and methods. AI is trained on existing information, so when a foreign threat occurs, the AI can have trouble recognizing and suppressing it, which Jenna Kerwin explains “can lead to privacy invasions, as sensitive information might be accessed or misused,” (Kerwin) leading to the corporation being blamed and negatively affecting their reputation. Bias can also be a problem in AI security, as false positives can lead to unfair treatment. This sometimes happens because artificial intelligence “notoriously lacks human judgment and creativity,” (Kerwin) and cannot tell the difference between normal human activities and malicious ones. Jenna Kerwin explains that the expenses of “special hardware, software, professional expertise, and training” (Kerwin) for AI incorporation are very high, overwhelmingly for smaller companies and startups. Employees must be retrained to work with the AI, or new employees will need to be hired, causing more expenses, time, and effort for the company. This makes small businesses a target for potential cyberattacks because they have simpler and more traditional systems that are easy to bypass with AI. Another limitation is that even though AI can run 24/7, it still needs human monitoring, meaning that the effectiveness of the system could depend on the time of day or other factors. These are exploits that scientists hope will be solved with more advanced AI that will require less supervision. A more advanced AI will enable organizations more time to work and spend less time monitoring, resulting in higher profits. New computing technologies like biocomputing could advance AI to make it faster and more flexible. Quantum computing will not arrive for a while, but AI and cybersecurity could be revolutionized by it.

Another concern in the world of AI is job displacement, as “Automating cybersecurity tasks through AI may also result in job displacement for cybersecurity professionals.” (Kerwin) Certain positions such as software designer and programmer could also end up becoming obsolete due to AI being the more efficient and effective worker. Because of these issues, “Some critics of AI-based solutions argue that technology could rapidly replace human workers, putting many IT professionals out of jobs.” (Kerwin) However, AI also creates jobs, for example the growing need for prompt engineers and AI specialists. Some experts argue that AI could lead to overreliance on AI, leaving items that need to be monitored by a human unattended. Additionally, the use of AI could allow a false belief that the AI will take care of everything on its own, meaning the AI goes unmaintained for extended periods of time. “Deepfakes and AI-generated misinformation blur the line between real and fake,” (Ghana) eroding trust between recipients in transactions and emails. Another concern is if the AI were to block an important or critical file from coming through, who would be to blame: the company or the AI? It is a prominent issue that is still being addressed by organizations and society. A growing problem in the cybersecurity world is the “AI arms race.” Both companies and cybercriminals are trying to get the best technology, leading to higher expenses for both sides. When AI is used for military applications, it brings concerns about whether we can trust the AI will not hack the system itself, especially as AI is integrated into unmanned vehicles and drones for warfare and fighting purposes. Giving the AI so much power over bombs and drones is a scary idea, which is why so many people oppose it. However, if we keep the parameters tight and have human supervision of it at all times, we have nothing to be afraid of.

Conclusion

AI is a tool that has become an integral part of the lives of millions, and a necessity for companies and countries alike to keep their data safe. As cybercriminals advance their attacks, organizations and nations must keep pace, or risk losing everything. It is exciting and dangerous as new technology advances and grows. In the coming era of quantum computing, dangers will become even more prominent, and organizations will have to work harder to keep up with the super-speed processing power of a quantum computer. More near in the future, biocomputing combined with AI could aid cybercriminals and organizations alike and lead to more flexible and adaptable algorithms. Overall, AI is a helpful tool with an incredible future

Works Cited

Kerwin, Jenna. "What Is the Role of AI in Cybersecurity?" The Role of Artificial   Intelligence (AI) in Cybersecurity, Excelsior University, 1 Aug. 2024,

 https://www.excelsior.edu/article/ai-in-cybersecurity/. 8 Nov. 2025

Meraj A., et al. "The Impact and Limitations of Artificial Intelligence in Cybersecurity: A Literature Review." The Impact and Limitations of Artificial Intelligence in Cybersecurity: A Literature Review by Meraj Farheen Ansari, Bibhu Dash, Pawankumar Sharma, Nikhitha Yathiraju :: SSRN, SSRN, 13 Jan. 2023, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4323317. 8 Nov. 2025

Sophos. "The State of Ransomware 2025." sophos-state-of-ransomware-2025.pdf, Sophos Ltd., 25 June 2025, https://assets.sophos.com/X24WTUEQ/at/9brgj5n44hqvgsp5f5bqcps/sophos-state-of-ansomware-2025.pdf. 8 Nov. 2025.

"AI in Cybersecurity: Key Benefits, Defense Strategies, & Future Trends." Artificial Intelligence (AI) in Cybersecurity: The Future of Threat Defense, Fortinet, Unknown Publication Date, https://www.fortinet.com/resources/cyberglossary/artificial-intelligence-in-cybersecurity. 8 Nov. 2025.

Boulton, Clint. "Human-In-The-Loop: Generative AI’s Rise Requires Hands-On Employees." Human-In-The-Loop: Generative AI’s Rise Requires Hands-On Employees, Forbes, 15 May 2024, https://www.forbes.com/sites/delltechnologies/2024/05/15/human-in-the-loop-generative-ais-rise-requires-hands-on-employees/. 8 Nov. 2025.

Varadajarin, Vivek. "Ethical and Adversarial Risks of Generative AI in Military Cyber Tools." Ethical and Adversarial Risks of Generative AI in Military Cyber Tools by Vivek Varadarajan :: SSRN, SSRN, 13 June 2025, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5287923. 8 Nov 2025.

"The Dark Side of AI in Cybersecurity: AI risks." Business and Financial Times (Ghana), 19 Aug 2025, https://advance-lexis-com.ezproxy.ohiolink.edu/document/?pdmfid=1519360&crid=09130a01-f260-44ab-85ea-0ddb55e4f429&pddocfullpath=%2Fshared%2Fdocument%2Fnews%2Furn%3AcontentItem%3A6GHY-W8M3-S7MR-D2KG-00000-00&pdcontentcomponentid=427903&pdteaserkey=sr1&pditab=allpods&ecomp=hc-yk&earg=sr1&prid=003f3960-0e72-4f58-bc81-853de7f4e554. 8 Nov 2025.

"The Future of Cybersecurity: AI-Powered Innovations Redefining Digital Defense.” Cyber Security Monitor Worldwide, 12 Apr. 2025, https://advance-lexis-com.ezproxy.ohiolink.edu/document/?pdmfid=1519360&crid=4c51e660-fb6f-4c0b-b061-bdbe48924328&pddocfullpath=%2Fshared%2Fdocument%2Fnews%2Furn%3AcontentItem%3A6FJF-0843-RS27-V3N3-00000-00&pdcontentcomponentid=476569&pdteaserkey=sr24&pditab=allpods&ecomp=hc-yk&earg=sr24&prid=d84fcda4-408f-400d-9ca5-05d4e1359ca6. 8 Nov. 2025.

(There was another source, but it could not be accessed when I tried to cite it)