artificial intelligence and cybercrime

Artificial Intelligence and Cybercrime: Keeping Up With a Rapidly Changing Threat

Artificial Intelligence and Cybercrime: Keeping Up With a Rapidly Changing Threat

By

Alphonso Nesbitt

CRJ 233 Cyber Crimes and the Law

10/24/26

EVALUATING VULNERABILITIES AND AI-DRIVEN FRAUD IN 55+ COMMUNITIES 3

Abstract

The use of artificial intelligence in online crimes has created big problems for the justice system. It's now harder to figure out who's behind a crime, verify evidence, and enforce laws. With AI tools making it easier to commit complex cyber crimes, investigators are struggling to keep up. They have a hard time verifying fake media, tracking down automated attacks, and meeting the standard of reasonable doubt. This study looks at how machine learning, digital evidence, and criminal law intersect. It shows how AI tools make it easier for people to hack into networks, cause problems across connected systems, and make judges doubt the authenticity of digital evidence. The study also highlights the lack of technical skills in investigative agencies and the weaknesses in old cybercrime laws, like the Computer Fraud and Abuse Act. To fix these problems, we need to update federal laws to clearly define and punish crimes like synthetic deception, algorithmic obstruction, and automated fraud. At the same time, justice agencies need to implement standardized frameworks for tracing digital evidence and securing prosecutions in a rapidly changing threat landscape. This will help them build stronger cases and keep up with the evolving nature of cyber crimes.

Artificial intelligence has significantly changed cybercrime, creating new challenges for investigators trying to identify, track, and stop digital attacks. Until recently, major cyber intrusions required extensive technical expertise. Attackers often had to develop their own malicious code, spend weeks or months probing systems for weaknesses, and carefully design targeted phishing campaigns. Today, powerful automated tools are widely available, allowing users to locate vulnerabilities, create convincing fake media, and launch phishing campaigns with minimal effort. This shift strains criminal justice institutions that were designed for an analog era. When cyber incidents involve AI content or automated attack networks, investigators often face major challenges in determining responsibility, verifying that digital evidence has not been altered, and proving guilt beyond a reasonable doubt. Basic security measures, such as employee training and routine network monitoring, can help prevent common threats, but they often fall short against sophisticated AI-driven attacks. Protecting the integrity of criminal proceedings in the age of AI will require updated cybercrime laws, stronger safeguards against synthetic deception, and uniform forensic protocols that allow investigators to work effectively across jurisdictions.

A major reason for this disruption is that advanced technologies are now widely accessible. Serious network intrusions used to be the exclusive signature of nation-state units or well-funded syndicates that maintained dedicated developer groups. Criminological findings show that common criminal motives, primarily extortion and illicit financial gain, have converged with consumer-grade machine learning utilities (Taylor et al., 2024). That dynamic collapses the historical barrier between opportunistic fraud and sophisticated enterprise-scale threats (Taylor et al., 2024). Ambitious offenders no longer require years of formal computer science training, now they can launch complex intrusion campaigns simply by feeding natural language prompts into publicly accessible software and automation scripts.

The fallout spreads far beyond isolated operational outages or immediate financial losses. As automated scripts strike system perimeters around the clock, single points of compromise frequently trigger cascading disruptions across interconnected civic and commercial infrastructure (Whyte, 2020). Treating this dynamic merely as a matter of computational speed overlooks how threat models have evolved (Whyte, 2023). Attackers routinely exploit artificial intelligence to manipulate social trust, mass-produce convincing synthetic personas, and spread coordinated disinformation across public channels (Whyte, 2023). Security teams often struggle to contain these attacks because automated tools can mimic routine administrative behavior, helping them evade detection and operate unnoticed within corporate networks.

These challenges don't stop with the investigation. They often carry into the courtroom, where existing rules for evaluating evidence were not designed to address deep fakes, videos, and other forms of digital deception. For decades, courts granted electronic records a functional presumption of authenticity as long as prosecutors established an unbroken chain of custody. Synthetic media engines dismantle that legal foundation. As deep learning models output photorealistic documents, audio clips, and surveillance footage, judicial and public confidence in digital proof is eroding (Dwork & Minow, 2022). Defense counsel can now plausibly argue that genuine security footage, wiretap recordings, email headers, or server logs are algorithmic fabrications, introducing reasonable doubt around completely untampered exhibits (Dwork & Minow, 2022). This creates a serious problem for prosecutors because people can claim that real digital evidence was generated or altered by AI, making it harder to convince juries that the evidence is real.

Compounding an already volatile situation, many investigative agencies lack the personnel and technical expertise needed to handle AI-driven cyberattacks. These cases often require specialists who can analyze complex digital evidence, spot subtle signs of malicious activity, and determine how an attack moved through multiple systems. Yet public safety and criminal justice institutions continue to face a stark shortfall of specialized cyber competencies (Demchak, 2026). Detectives and state investigators rarely receive the technical instruction needed to interpret algorithmic footprints, map rotating proxy layers, or reliably anchor autonomous bot activity to an individual keyboard (Demchak, 2026). When malicious software is designed to blend in with normal network activity and conceal its origin across multiple countries, investigators often struggle to prove responsibility and gather enough evidence to support criminal charges.

Addressing this institutional gridlock requires moving away from piecemeal statutory amendments and toward structurally adaptive legal regimes. Core cyber statutes, particularly the federal Computer Fraud and Abuse Act alongside state-level equivalents, were constructed around the physical concept of breaking through an unauthorized digital perimeter. They were never designed to penalize modern algorithmic abuses, such as voice cloning used in grand larceny, synthetic biometric spoofing, or automated record tampering. Rather than stretching outdated statutes to cover machine learning realities, lawmakers need flexible architectures built to govern autonomous systems and adapt alongside technological change (Cuéllar & Huq, 2022).

References

Cuéllar, M. F., & Huq, A. Z. (2022). Artificially intelligent regulation. Daedalus, 151(2), 335–347. https://doi.org/10.1162/daed_a_01920

Demchak, C. C. (2026). Cyberattack-resilient artificial intelligence in navies: Solving the skills-need gap. Naval War College Review, 79(2), 47–72.

Dwork, C., & Minow, M. (2022). Distrust of artificial intelligence: Sources & responses from computer science & law. Daedalus, 151(2), 309–321. https://doi.org/10.1162/daed_a_01918

Taylor, R. W., Fritsch, E. J., Liederbach, J., Saylor, M. R., & Tafoya, W. L. (2024). Cyber crime and cyber terrorism (5th ed.). Pearson.

Whyte, C. (2020). Poison, persistence, and cascade effects: AI and cyber conflict. Strategic Studies Quarterly, 14(4), 18–46. https://www.jstor.org/stable/26956151

Whyte, C. (2023). Beyond "bigger, faster, better": Assessing thinking about artificial intelligence and cyber conflict. The Cyber Defense Review, 8(3), 135–150.