HOUSTON, Texas (KTRK) — With midterm early voting just days away, AI companies are warning people to watch for AI-generated content created to mislead and confuse voters.
“It’s advanced so much, so fast that you have to be very critical of everything,” Andy Perez, a Harris County voter, said.
Jacqlyn Ocumpaugh, a University of Houston education professor who studies AI, said that as the technology improves, it is less about looking for mistakes and more about looking for content that seems too perfect.
“Particularly if it’s a human being, the lighting is really studio quality in a lot of these, and a lot of people don’t have the benefit of a cameraman following them around,” Ocumpaugh said.
There are no federal laws in the United States that require AI content to be labeled, something voters said they would like to see change.
“It should be watermarked, so you can have your choice if you want to believe it or not. I think when it comes to that, the choice is what matters about the whole situation,” said Jnee Adams, a Harris County voter.
Some AI companies have created tools to help. OpenAI, the parent company of ChatGPT, has built a tool that lets users drop in a photo to find out if it was made with AI.
Anthropic, the company behind Claude, said it will soon use watermarking to comply with the new European Union AI Act, which requires AI systems operating in the EU to mark when content is AI-generated.
But experts caution that even a photo not made with AI can still carry false information.
“It’s like a hammer. When you are using it to build a house, it’s a great tool; if you use it to knock someone upside the head, it’s not a nice tool anymore,” Ocumpaugh said.
Copyright © 2026 KTRK-TV. All Rights Reserved.