Artificial intelligence is changing the way fraud works—and who it targets. What was once easy to spot through bad grammar, generic messages, or obvious scams has evolved into something far more convincing. Today’s AI-driven fraud doesn’t just imitate institutions. It imitates people. From cloned voices to hyper-personalized messages, these attacks are becoming harder to detect and more emotionally manipulative. And while much of the conversation has focused on corporate risk, the reality is closer to home: individuals and local communities are increasingly on the front lines. To better understand how this shift is playing out, and what people can do about it, we spoke with Brian Peret, Director of CodeBoxx Academy.

When Fraud Starts to Sound Like Someone You Love

AI-driven fraud has moved beyond obvious warning signs. Instead of casting wide, generic nets, it now targets individuals with precision. “AI-driven fraud has moved past the era of obvious red flags,” Peret explains. “Today, it targets the most personal aspects of our lives. It targets our voices and our faces.” One example is the evolution of so-called “grandparent scams.” Using voice cloning technology, scammers can now replicate the tone and cadence of a loved one, creating urgent and emotionally charged scenarios that feel real enough to bypass skepticism. At the same time, phishing attacks have become deeply personalized. AI systems can scrape publicly available data to craft messages that reference specific jobs, local events, or even recent purchases, details that make fraudulent outreach feel legitimate. Another emerging threat is recruitment fraud. Fake job postings, sometimes referred to as “ghost jobs,” are used to lure applicants into sharing sensitive personal and financial information through AI-generated interviews.

Why AI Makes Fraud Harder to Detect

For years, people relied on instinct to spot scams. Looking for typos, awkward phrasing, or generic greetings. AI has erased many of those signals. “The primary challenge is that AI has effectively neutralized our biological error detectors,” Peret says. Instead of poorly written messages, individuals now face “linguistically perfect, hyper-personalized narratives” that mirror real communication styles. These messages don’t just pass technical filters—they bypass human intuition. More concerning is the emotional realism AI introduces. Voice cloning and deepfake video can trigger immediate reactions, especially in high-pressure situations. “When a scammer can clone a grandchild’s voice or mimic a trusted colleague’s face, they trigger an immediate ‘fight or flight’ response,” Peret explains. “This psychological hijacking makes it incredibly difficult for an individual to pause and apply logic.” Because these attacks can be deployed at scale, communities are now dealing with a constant flow of high-quality, emotionally convincing threats.

The Case for AI Literacy at the Community Level

As these risks grow, the response is shifting beyond traditional cybersecurity advice. Increasingly, experts point to education as a critical line of defense. “Education and workforce development programs act as the essential translation layer between complex emerging technology and practical, everyday defense,” Peret says. Rather than focusing only on what to avoid, these programs aim to build a deeper understanding of how AI works—and how it can be misused. That includes adopting what Peret describes as a “Zero Trust mentality,” where individuals treat digital interactions as unverified until confirmed through a secondary channel. It also means demystifying the technology itself. “When a local training center demonstrates exactly how a voice is cloned or how a deepfake is rendered, the magic of the scam disappears,” he says. “It’s replaced by informed skepticism.” By embedding this kind of literacy into workforce development, communities can create what Peret calls a “human firewall”—a collective layer of awareness that is harder to penetrate than any single piece of software.

New Warning Signs in the Age of AI

Even as AI improves the quality of deception, it can still leave subtle clues. In audio scams, listeners may notice unnatural pacing—such as a lack of breathing pauses or a tone that feels unusually consistent. In video, small visual inconsistencies can appear, including blurring around the mouth, flickering edges, or mismatched lighting. But the most reliable signals are often behavioral. Attackers may struggle with unexpected questions or avoid direct verification. And urgency remains a key tactic. “If a caller creates an intense sense of urgency but fails to verify their identity through a known method, it is almost certainly a machine behind the mask,” Peret notes.

A Shift That Requires a Different Kind of Defense

AI is not just making fraud more advanced. It is making it more human. And that changes the nature of defense. Technical safeguards still matter, but they are no longer enough on their own. Awareness, education, and critical thinking are becoming just as important. As these tools continue to evolve, the challenge will not be keeping up with every new tactic. It will be building the kind of understanding that allows people to recognize when something feels real—but isn’t. Because in this new landscape, the most dangerous scams are not the ones that look suspicious. They’re the ones that don’t.