The Rise of Cybercrime Threats: How AI Is Changing Hacking Tactics

Artificial intelligence (AI) technology has become a powerful weapon in the arsenal of modern cybercriminals. This shift creates a much more complex threat landscape, with attacks that are faster, cheaper to execute, and far more difficult to detect. Experts warn that cybercrime has entered a new era where scale and reach are the main driving factors.

Rapidly Growing Cybercrime Through AI Automation

Leading AI companies such as Anthropic, OpenAI, and Google have reported significant increases in the use of their platforms for malicious activities. Criminals leverage this technology to design complex spear-phishing operations, develop sophisticated malware, and carry out multi-stage digital attacks.

One of the most striking changes is the emergence of deepfake audio and video used to deceive employees. Scammers can now create fake multimedia content that mimics company leaders, forcing employees to reveal sensitive information or transfer funds unsuspectingly.

Alice Marwick, a researcher at Data & Society, explained this phenomenon to The Wall Street Journal with a simple yet impactful statement: “The real change is in scale and scope. Fraud is bigger, more targeted, and more convincing.”

AI Phishing Messages Dominate the Threat Landscape

Researcher Brian Singer from Carnegie Mellon University, who specializes in the use of large language models in cyberattacks, estimates that half to three-quarters of all spam and phishing messages worldwide now originate from AI systems.

The credibility of these messages reaches a new level. AI systems trained on corporate communication styles can generate thousands of messages that sound natural and consistent with the target organization’s writing patterns. These messages:

• Mimic how executives write with high accuracy
• Refer to recent news from public records
• Eliminate language errors that previously exposed international scam attempts

John Hultquist, head analyst at Google Threat Intelligence Group, describes this shift as “credibility on a large scale.”

Data-Driven Targeting: Identifying Vulnerable Victims

Malicious actors are becoming more sophisticated in selecting their targets. They use AI to sift through social media and identify individuals facing significant life difficulties—divorce, family death, job loss—that make them more vulnerable to romance scams, investment fraud, or fake job offers.

This data-driven approach dramatically increases scam conversion rates.

Dark Web Lowers Barriers to Cybercrime

The underground ecosystem has evolved into a distribution hub for criminal AI services. These platforms offer hacking tools at affordable prices—starting from $90 per month—complete with tiered customer support.

These services operate under names like WormGPT, FraudGPT, and DarkGPT. They enable:

• Creation of malware without manual coding
• Automated large-scale phishing campaigns
• Structured hacking technique tutorials for beginners

Nicolas Christin, head of the software and systems community division at Carnegie Mellon, explains this business model: “Developers sell subscriptions to attack platforms with tiered pricing and full customer support.”

Margaret Cunningham, vice president of AI security and strategy at Darktrace, simplifies the skill barrier: “You don’t need to know how to code, just where to find the tools.”

The latest development called “vibe-coding” or “vibe-hacking” allows aspiring criminals to use AI to create their own malicious programs instead of buying from the dark web. Anthropic revealed earlier this year that they stopped several attempts to use their Claude AI to create ransomware by “criminals with minimal technical skills.”

Industrialization of Cybercrime Operations

Cybercrime has operated like a specialized business ecosystem for years. Typical ransomware operations involve division of labor:

Access brokers break into corporate networks and sell access to the next group.

Penetration teams move through systems to steal valuable data.

Ransomware service providers deploy malware, manage negotiations, and distribute profits.

AI has revolutionized this model by increasing speed, automation, and scalability. Tasks that previously required specialized technical skills can now run automatically, allowing criminal organizations to operate with fewer personnel, lower risk, and higher profits.

Christin makes a fitting analogy: “Think of this as the next industrial revolution. AI boosts productivity without requiring more skilled labor.”

Can AI Fully Conduct Attacks on Its Own?

This critical question has yet to be answered affirmatively. Experts compare the situation to autonomous vehicle development. Five percent of automation has been achieved, but the last five percent that would enable fully independent operation is still far off.

Researchers are testing AI hacking capabilities in laboratory environments. A team at Carnegie Mellon, supported by Anthropic, successfully simulated a well-known Equifax data breach using early AI—what Singer calls a “big leap” in attack automation capabilities.

The Role of AI in Defense: Any Hope?

While criminals leverage AI for malicious purposes, the same technology offers powerful defense tools. Anthropic and OpenAI are developing systems that can continuously scan software code for vulnerabilities before criminals find them.

Recent AI programs developed by Stanford researchers outperform some human security testers in identifying network vulnerability issues.

However, Hultquist emphasizes that AI is not a magic pill. Organizations must focus on building resilient networks—systems that continue to operate even when under attack. Absolute prevention is impossible; resilience is the real goal.

Organizations and individuals who understand this evolution in cybercrime are already one step ahead in their defenses.