Researchers warn AI can make people’s beliefs easier to influence

In short: Researchers are increasingly warning that AI can make people’s perception and beliefs easier to shape, mainly through fake media, misinformation, and repeated AI driven feedback.

What's going on

Deepfakes and other synthetic media, meaning audio, images, or video made by AI, are getting better and cheaper to produce. This can make it harder for people to trust what they see and hear. It can also create a general sense that “anything could be fake,” which can weaken confidence even in real evidence.

Researchers also point to human and AI feedback loops. This is when people use AI outputs to make decisions, then those decisions and reactions feed back into what the AI shows next. Like repeating a rumor in a small town, even a small mistake or bias at the start can grow over time, and studies suggest this can increase human error.

Another finding is that AI systems can change how people frame the world without “knowing” the world. They do it by filtering and summarizing information based on the data they get, like a pair of tinted glasses that changes what stands out. A 2026 study in AI & Society also found that simple claims about AI being “sentient” (able to feel or have experiences) can measurably shift how much people trust it and whether they judge it as conscious.

What to watch

The practical risk is not that facts stop existing. It is that more people may rely on AI answers that can be wrong, or believe convincing impersonations, or get pulled into information bubbles where errors reinforce themselves. Expect more pressure on platforms and AI tool makers to label synthetic media, slow the spread of false content, and help users verify what is real.

Source: NYTimes