Dr. Tina Nikoukhah

How did you first become interested in digital forensics?

If I broaden the field to multimedia security, my first encounter was during a school project when I was 19. I discovered digital watermarking — the idea that you can embed hidden information in an image and protect it throughout its lifetime. I was fascinated by how a mathematical formula could add content to a picture without altering its appearance.

Was there a defining moment that made you realize this was your passion?

I was already passionate about image forensics when I started my PhD since it felt like a natural extension of that early experience.

But one moment really stood out. In 2018, I came across a viral image of a young woman, manipulated to make it seem like she was tearing up the American Constitution. At the time, I was working on my first method for detecting local manipulations based on compression traces. Out of curiosity, I ran my method on that image — and it detected the manipulation. Even better, I could explain why: the resampling and warping operations had disrupted the original compression traces, revealing the forgery.

That’s when I realized my research wasn’t just about solving theoretical problems or producing results on carefully curated datasets. It had a real-world impact, and I knew I wanted to keep going.

How do you envision your research evolving in the next few years?

I think we’ve reached a point where manipulated or generated content can appear so realistic that the human eye alone can’t detect it. Some platforms are starting to integrate responsible practices, adding signatures like C2PA or watermarking to improve traceability. However, these markers can disappear when content is reshared across platforms.

I believe the future should focus on both regulating and standardizing these practices to ensure such markers become widespread and persistent. This would make content authenticity easier to verify. In parallel, our research will continue to focus on developing robust detection methods that adapt to emerging tools, especially for cases where no signature is present. 

How could your work change how people interact with digital media?

I hope my work raises awareness that not everything we see, hear, or watch is authentic. Just as we wouldn’t mistake a drawing for reality, people should learn to question digital content rather than blindly trust it.

What’s your unique approach or philosophy toward research?

 My research approach — and I’m surely not the only one — is to keep things simple. The simplest solution might not always make it to top conferences, but if it works on real-world data and is explainable, that’s what truly matters.

What’s a misconception people have about your research area?

Many believe media forensics can be solved by simply feeding real and fake data into the biggest deep learning models. But even if that worked, it wouldn’t explain why a piece of content is manipulated, and that’s often the most critical part.

What’s one piece of advice you’d give to aspiring researchers?

I encourage anyone interested in research to consider pursuing a PhD. It’s an incredible opportunity to develop scientific rigor and valuable skills that apply far beyond academia. It is more than just gaining expertise; it’s learning to think critically, solve problems, and manage uncertainty.