I was at a store the other day and spotted something that made me do a double-take: a 2025/2026 brochure for Downtown Ithaca—one of those business directory/event guide/map combo things. Nice production value. And right there on the cover, doing some serious heavy lifting, was one of my photos.

Well, half of my photo.

Was I credited? No. Paid? Of course not. Do I care? Not really—they pulled it from my Unsplash page, probably with no idea a local photographer took it. That’s how free stock photos work. I get why they cropped it too. The facades match the Downtown Ithaca logo perfectly, showcasing the buildings along our Commons pedestrian mall. Smart choice.

And here’s what they cropped out, and why it’s interesting.

Look closely at the full photo and you’ll spot people wearing KN95 masks. Not exactly the vibe you want to project in 2025. (Though I still see masks around town occasionally—not many, but some.)

My photo is from September 20, 2020. Peak pandemic. That group of people wouldn’t normally be standing there at all—it was a pandemic thing. Without them, I probably wouldn’t have taken the shot.

Fun historical detail number two: Waffle Frolic, the business in focus, closed years ago. It’s been two different businesses since then, neither worth stopping in if I’m being honest.

So here’s the paradox: In an age where AI can generate almost anything, we’re still relying on real photos of real environments. The pre-AI era images sitting on Unsplash, Pexels, and Pixabay are more powerful than ever. But should they be? Are they being scraped to train AI models anyway? When convenience and price are why people hit “generate,” should we be more protective of our “free” assets?

That said, there are a dozen photographers in town who could’ve produced a better, more current image for maybe $100. So did I steal their potential work by putting my photo on Unsplash for free? Am I worse than AI?

I used to ask myself the same thing about free fonts I’d release. Here’s what I learned: Free things get used because the people using them were never going to pay anyway. The market for free and the market for paid don’t really overlap. But how does LLM image generation fit into that?

I was curious if my photo could be replicated quickly, easily, and for free. I asked ChatGPT to create a “hyper-realistic photo of the Commons pedestrian mall area in Ithaca, New York.”

It’s… not great. But if you didn’t know the town and didn’t look too hard, maybe you could use it. I wouldn’t recommend it for a dozen reasons. And more and more, it seems like most people are fine with “good enough.” Which is disturbing. I tweaked the prompt to get closer to my photo’s angle.

Still…not quite. But there’s a growing number of folks who’ll look at this and say “close enough.”

I stopped putting photos on Unsplash last year. Even though I genuinely enjoy sharing free resources—especially ones rooted in real life. This post didn’t start as a rant about AI image generation, and I don’t want it to end that way either.

As real as my photo is, it’s also limited. It’s a record of a very specific moment in time that is complicated and messy and special. Really I think the way to stay authentic with images in general, is to keep taking them. The street you photograph today won’t look exactly the same tomorrow, and that’s a wonderful thing. A natural thing.