Imagine scrolling through your social media feed and seeing a video of a familiar face—a family member, a trusted colleague, maybe even a public figure—saying something they would never say. It looks real, sounds real, but it is not. This is the unsettling reality of deepfakes, synthetic media created to mimic a person’s likeness and put false words or actions into their mouth. Deepfakes have grown from a niche internet curiosity into a widespread public safety concern, with studies showing a 550% increase in deepfake content between 2019 and 2023. As artificial intelligence tools become more accessible and easier to use, the line between genuine and fabricated media has blurred dramatically. Scammers can now clone a voice with just a few seconds of audio, fraudsters can create convincing video calls from a single profile picture, and malicious actors can spread misinformation faster than fact-checkers can correct it. For everyday people, this creates a terrifying question: If we can no longer trust our own eyes and ears, how do we know what is real? This is precisely the problem that the Michigan Deepfake Resilience Hub, or M-DRH, is trying to solve. Led by Khalid Malik, a professor of computer science and director of cybersecurity at the University of Michigan-Flint’s College of Innovation & Technology, the tri-campus initiative is working to make deepfake detection tools accessible to everyone—not just governments, courts, and tech experts, but the average citizen who just wants to know whether a suspicious video is genuine or manipulated.
At the heart of M-DRH is a simple but powerful recognition: existing detection technologies are not reaching the people who need them most. In an interview with The Michigan Daily, Khalid Malik explained that while detection tools exist, they are often fragmented, highly technical, and inaccessible to the public. “The problem is: Who will verify that this is real or fake?” he said. “The technology piece is missing and we are trying to fully fulfill this gap. We are developing detection as a service. We are already helping the governments and the courts, but this we are making for the common people. So we want to use the best practices of human-centered design.” This vision has transformed M-DRH from an academic project into a community-focused resource. The initiative recently received a significant boost when it was selected as one of 37 projects out of 450 proposals to receive funding from the University of Michigan’s first round of Strategic Initiative Fund awards. The SIF is a $1 billion investment distributed over ten years, designed to support high-value, high-impact initiatives across all three UM campuses. For M-DRH, this funding means the opportunity to build a public verification portal where anyone can submit suspicious media and receive analysis, train more than 1,000 educators as “Digital Integrity Ambassadors,” and develop policy guidance to help protect Michigan residents from AI-generated fraud and misinformation. The initiative is not just about catching fakes—it is about building a culture of critical thinking and digital resilience, so that people can navigate an increasingly deceptive digital world with confidence rather than fear.
The challenge is daunting, in part because deepfake technology is advancing at a breathtaking pace. Hafiz Malik, a professor of electrical and computer engineering at UM-Dearborn and a member of the M-DRH project team, described the situation as “uncontrollable” in an interview with The Daily. He pointed out that there is currently no single platform where someone can go to get all their questions answered—whether a piece of text, an audio recording, or a story was generated by AI. “We are trying to put together these resources and expertise so that you can find answers to these questions under one roof,” he said. The goal is to create a centralized, user-friendly resource that draws on the combined knowledge of researchers across the University of Michigan system. But even as M-DRH works to build this unified platform, the researchers are painfully aware that detection is an arms race. Bad actors are constantly learning how detection algorithms work and finding ways to circumvent them. Hafiz Malik noted that for audio, in particular, human detection is nearly impossible. “There is no chance that a human can detect it,” he said. “Bad actors know how the technology works for detection, and they know how to get around it. If you have pristine synthetic content generation, detection is solid. That is a big challenge, and having unified platform availability for that is also a challenge.” In other words, the very perfection of deepfake technology is what makes it so dangerous—and why automated, accessible detection tools are no longer a luxury, but a necessity.
Beyond the technical hurdles, there is a deeply human dimension to the deepfake crisis. This is not just about politicians being misquoted or celebrities being placed in embarrassing situations. It is about real people being scammed out of their life savings, families being torn apart by fabricated evidence, and victims of non-consensual content suffering profound emotional and reputational harm. Shirley Giraldo, a Business graduate student and co-president of the AI and Emerging Tech Club, spoke to The Daily about the importance of M-DRH’s public service mission. “This Hub is being developed to help citizens, educators and public institutions,” she said. “I think it is providing a crucial public service that extends beyond college students. I think there are a variety of risks when it comes to deepfakes, but perhaps being scammed by what appear to be ‘friends’ would pose the biggest and likeliest risk to an individual.” Giraldo’s point is chillingly practical. Imagine receiving a voicemail from a loved one in distress, asking you to wire money immediately. The voice sounds exactly like them—because it is their voice, cloned by AI. Or imagine a video call from a colleague requesting a confidential password, with their face and mannerisms perfectly replicated. These are not hypothetical scenarios; they are already happening. According to the FBI, deepfake-related fraud reports have surged in recent years, with criminals using synthetic media to impersonate executives, romantic partners, and even family members. The financial and emotional toll on victims is devastating. M-DRH’s public verification portal is designed to be a first line of defense, giving ordinary people a place to check suspicious content before they act on it. In a world where trust is increasingly fragile, having a reliable, accessible way to verify media can mean the difference between safety and victimization.
Of course, no technological tool can solve this problem alone. Hafiz Malik emphasized that point strongly: “This problem cannot be solved through technology only,” he said. “It can be solved through a multi-pronged approach: technology, regulation, enforcement and public awareness outreach. So, if you put these things together, you can mitigate the damages (deepfakes are) going to impose on the society at large.” This holistic philosophy is woven into the fabric of M-DRH. On the technology side, the initiative is developing sophisticated detection algorithms and a user-friendly verification platform. On the regulatory side, it is working to create policy guidance that lawmakers and institutions can use to combat AI-generated fraud—laws that would hold creators of malicious deepfakes accountable and protect victims. On the enforcement side, M-DRH is partnering with government agencies and courts to ensure that law enforcement has the tools and training needed to investigate deepfake crimes. And on the public awareness side, the initiative is investing heavily in education, training over 1,000 Digital Integrity Ambassadors who will then teach others in their communities how to spot suspicious media and protect themselves. This multi-pronged approach recognizes that deepfakes are not just a technical problem but a social one. Technology can help us detect fakes, but only education and awareness can help us question what we see, verify sources, and think critically before sharing or acting on information. By addressing the issue from every angle, M-DRH hopes to create a safety net that catches deepfakes before they cause harm—and builds long-term resilience in communities across Michigan and beyond.
Perhaps the most reassuring aspect of the Michigan Deepfake Resilience Hub is that it is not starting from scratch. The initiative began in 2017, long before deepfakes became a household word, and has spent years developing relationships with government agencies, courts, and research partners. The recent SIF grant is not a launch so much as an expansion—a chance to scale up proven approaches and make them available to the public in a way that was never possible before. For Khalid Malik and his colleagues, this is a deeply personal mission. They have seen the damage that deepfakes can cause, and they know that the next victim could be anyone: a retiree targeted by a voice-cloning scam, a student whose image is used without consent, a small business owner whose reputation is destroyed by a fabricated video. M-DRH is their answer to that threat—a place where technology meets compassion, where cutting-edge research is translated into practical tools for everyday people. In the end, the goal is not just to detect deepfakes but to restore a sense of safety and trust in the digital world. As artificial intelligence continues to evolve, the line between real and fake will only become harder to draw. But with initiatives like M-DRH, communities are being equipped with the knowledge, tools, and support they need to navigate that uncertain terrain. The road ahead is difficult, and no single solution will be enough. But by bringing together researchers, educators, policymakers, and ordinary citizens, the Michigan Deepfake Resilience Hub is proving that we are not powerless in the face of AI-driven deception. We are learning to fight back—together, with eyes wide open and a shared commitment to protecting what is real.