A photographer said their own image was rated as 89% likely to be AI-generated, prompting frustration over detection tools and questions about how they assess photographs processed by smartphone software.
Here's what to know
In a Reddit post, the photographer said they had taken the image with an iPhone 16e using night mode, describing the feature as "just a longer exposure."
"These tools for detecting AI images and text are mildly infuriating," the original poster wrote, sharing screenshots of DeepAI's AI detection results for two different images — one was identified as 89% AI, while the other was 86% AI.
Several commenters explored whether the phone's software might explain the result. One suggested that newer phones may use AI-assisted cleanup or upscaling unless those features are disabled.
Another questioned whether the detector was identifying an artificial creation or reacting to software processing applied to a camera image.
"There's a difference between images generated by AI and AI being used in post processing, and you're right that the latter is done not just with phones but even professional photographers using AI for denoising," they wrote. "But the alert in OP's screenshots specifically states likelihood that the images were 'created by AI.'"
TCD Picks » Quince Spotlight
💡These best-sellers from Quince deliver affordable, sustainable luxury for all
More background
Other commenters shared similar experiences with detection tools. One commenter said the same site had flagged both a photograph they took of the moon and a 3D render they made as probably AI-generated.
"AI detection services are unreliable in general except for SynthID checks," one commenter warned.
Another debated that claim, "Well, I wouldn't exactly say reliable. You can still remove a SynthID. But you can't get a false positive. So in that sense, yeah. It's reliable."
Another Redditor added, "That's all right. I just ran some of the Bible through an AI detector and it turns out the Bible is 78% AI. Who knew?"
What can be done?
Some suggested checking for SynthID and similar watermarking systems rather than relying on tools that estimate whether an image looks AI-generated.
"This is what I hate about the AI checkers, they're terrible at what they actually do, yet people will use them like they can never fail," one commenter wrote.
Where can I learn more?
Similar questions about image authenticity have surfaced in rental listings and damage claims.
• In Australia, a rental photo flagged as AI-doctored actually showed a $5,000 Bondi mural.
• A developer found an invisible watermark on AI-made images in Microsoft Paint and Photos.
• Airbnb damage claims prompted questions about whether a toilet photo was AI-generated.
• A home shopper criticized AI staging after a purported dining room appeared atop a staircase.
• In Ireland, a DNG-linked agent apologized after AI-altered apartment photos drew public backlash.
Get TCD's free newsletters for easy tips, smart advice, and a chance to earn $5,000 toward home upgrades. To see more stories like this one, change your Google preferences here.
