It’s been about a month since we opened user registration on WasItAI, our AI image detector, and we wanted to share a quick update on what we’ve been working on behind the scenes.
For anyone new here: WasItAI is an AI image detector that helps identify whether an image was likely created by AI or by a human. Since opening registration, we’ve received a huge amount of feedback from artists, photographers, researchers, and everyday users.
First of all – thank you.
Many of the improvements we made over the last month came directly from community feedback.
Making AI Image Detection Feel More Realistic
One of the biggest things we heard was that the AI image detector sometimes felt too confident in difficult or ambiguous cases.
We agreed.
Over the last few weeks, we adjusted predictions to feel more balanced and realistic, rather than always expressing complete certainty. AI image detection is rarely black-and-white, especially now that AI-generated and human-created images are becoming more visually similar every month.
Reducing False Positives
We also received many reports from artists, photographers, and creators whose real images were flagged as AI-generated.
Reducing those false positives became a major focus this month.
We put significant work into improving how the detector handles artwork, stylized images, renders, edited photos, and other difficult cases that can resemble AI-generated content.
We’ve already seen noticeable improvements, but we want to stay honest about something:
No AI image detector, including ours, will ever be perfect.
The detector may still flag some human-made images incorrectly. Some AI-generated images may still slip through. As image generation tools keep improving, the line between AI-generated and human-created content gets harder to draw.
We don’t want to pretend otherwise or oversell what the system can do.
And if you get a wrong result, report it. We genuinely take those cases into account when improving the detector.
We’re Listening
What we can promise is that we’re actively improving WasItAI, listening closely to feedback, and continuing to work on the cases users care about most.
So if you’ve reported a weird prediction, sent examples, or shared feedback over the last month, thank you. Seriously. Many of these improvements happened because people took the time to help us improve the system.
We’ll keep refining the detector, reducing false positives, and improving the reliability of our AI image detection over time.
And we’re just getting started.
WasItAI R&D team