Artificial intelligence has woven itself so seamlessly into the fabric of modern life that we often forget it’s there. It’s in our pockets, dictating our text messages and choosing the music we listen to. It’s in the apps we use to edit photos and the tools we rely on for work. But with the recent explosion of generative AI—the kind that can create stunningly realistic images and videos from scratch—the line between what is real and what is fabricated has become troublingly blurred. This isn’t an abstract concern; it’s about the very trust we place in the images we see, the news we consume, and the people we elect to power. The question that now hangs over society is deceptively simple: in a world where a computer can make a politician appear anywhere, how can we ever be sure of what’s real? This is the challenge that News4JAX investigator John Asebes decided to tackle head-on in an experiment designed to test the very tools meant to answer that question.
To understand the effectiveness of the technology, Asebes started with a bit of playful curiosity. He asked ChatGPT to generate a believable photo of himself holding the NHL’s Stanley Cup—a scenario so obviously fanciful it seemed harmless enough to test the power of these new tools. But even to his own eye, something was subtly off. From the strange overlap of familiar facial features to a certain artificial smoothness, the image possessed an uncanny valley quality that made him suspect he was looking at something that wasn’t real. The most revealing part of the exercise, however, came next: he ran this obviously fabricated image through three different commercial AI-detection websites. One would think that if the same AI they were trained to catch generated the image, their algorithms could easily spot it. Instead, the results were a mess. One detector confidently stamped the image as “unlikely to be real,” while a second returned a judgment of “uncertain,” and a third flagged it as likely AI-generated. There was no consensus, no single voice of authority to give a definitive answer. As a test, Asebes also inflicted the same tools to a genuine, verified photo of him and a colleague, and even then, the detectors didn’t give it a clean bill of health. One of the statistically most “real” scores was only 85% sure, while the rest offered even more hedged percentages. The issue is that these detectors use probability, not certainty, and proof of this became the pattern. When it should be a black-and-white answer, these programs return a murky gray.
This inconsistency isn’t just a minor technical quibble; it carries massive weight when the subject goes beyond a fictional photo op. Consider the recent case involving Senator Mitch McConnell. After a serious fall that led to hospitalization, the senator disappeared from the public eye for an extended period. When his office released official photographs of him in rehab, the internet erupted with speculation. Were these images a way of showing the world he was healing, or were they AI-generated fabrications design to hide a more concerning decline in his health? In times like this, someone has to trust medical records, political statements, or the reputation of the lawmaker’s office. When News4JAX tested these official photos, the results were just as split, just as hesitant, and just as unreliable as the Stanley Cup test. Some detectors said they were likely real; another said uncertain, and a different one outright flagged it as probably fake. This is where the rubber meets the dinner conversation in real life. We’re left relying on the reliability of what—or who—is disseminating these images. “You’re trusting whoever is pushing out that image, even if it is associated with an official account,” a source in the story said. “We already know, for instance, that politicians have put out AI-generated media, and sometimes they disclose that, other times they don’t.” It’s a terrifying assertion of the decline of institutions, when even the official account doesn’t trust the image from their own leader.
To strip away the abstract and talk about the moral, issue of it all, it was time for Doctor Josh Gellers, an AI expert, to weigh in. Asebes spoke with him about the scale of the change and the role we each play in navigating the chaotic new ecosystem of image-making. Gellers, who is the Dean of Professional and Lifelong Learning at the University of North Florida, offers a perspective that goes beyond the tools and goes into the matter of the societal impact. “We haven’t fully grappled with the magnitude of changes that this is going to cause,” Gellers said. As a researcher, he argues that we need to be actively thinking about AI as we use it in our professional and personal lives. His comment pointed to the fact that there’s no relenting on the rollout of generative AI. It’s being integrated into every sector—public, private, education, even religion—in record time, and there is no way to magically reverse it. It means our best line of defense isn’t the technological relay race of detector versus developer. It’s taking a step back to reflect on the ethics and morals of a world with generated content, both for ourselves and on how we shape the story—institutions and individuals alike need to be asking questions: where should we allow AI to interact with our lives and where should we draw a line?
The impact of this blurry boundary has already reached into real human lives. In Jacksonville, Florida, one’s life was altered forever the moment someone online created a fabricated, explicit nude photo of her and posted it for the world to see. Brooke Curry, the daughter of former Jacksonville Mayor Lenny Curry, became a reluctant activist overnight. She described the effect to News4JAX as an attack not just on her body and her innocence, but on the way that people saw her and the reputation she had built in her name. “You have to go through that to really understand the trauma it brings,” Brooke shared. “The hardest part was that people behind closed doors don’t understand the impact it has on you.” Her experience prompted her to take legislative action with a law firmly in Florida’s corner’s framework: Brooke’s Law, passed in 2025. Under this new statute, digital platforms are required to take down AI-generated sexually explicit images that are posted without the subject’s consent after the victim sends a valid request. It was a huge step in protecting individuals from the damaging potential of deepfakes—but it’s only one act of addressing a large and multifaceted problem. Florida has also created new, stricter definitions for generating AI child sexual abuse material, and made it illegal to use AI to create moveable political ads without a disclaimer. Yet there’s more on the table that was considered and lost: a bigger universal “AI Bill of Rights” passed through the state Senate but failed in the House, showing how complex and tricky it is to regulate such a shapeshifting, fast-changing tech.
For all of us trying to keep up in this shifting world, we have a simple charge: create a new critical mindset. When informciliars are increasingly believable, that it’s hard to know what to trust. The concerning news is that the technology isn’t slowing down. The experts all agree that it’s becoming smarter, faster, and more indistinguishable day after day. As Dr. Gellers points out, the key that not to check on AI, but to always keep our person to check. The concept of “personhood” is becoming a unique shield. We should think about the places where we do not want AI to be involved because they affect now much of our humanity we’re reserving. If something feels or looks off, if a image seems too convenient or too perfect, which is a cue to pause. When in doubt, think before you share. These tools are meant to help us, but left unchecked, they can potentially threaten livelihoods, reputations, and even our sense of self. In the span of just a few years, we have traded some of our “seeing is believing” for the tools to spread doubt. The answer is not to simply be alarmed, but to be aware. So, be curious, be skeptical, and critical with what you see online. The key to holding onto what’s real is not in any technology or detector—but in the human act of looking, questioning, and acting with care.