FAUXLENS

Detect DALL-E AI Images

Upload an image and find out whether OpenAI's DALL-E made it. FauxLens reads content provenance metadata where it survives and examines the picture itself where it does not, so a stripped or re-saved file can still be identified.

JPG PNG HEIC · MP4

Detects content from

  • OPENAI
  • MIDJOURNEY
  • Stability AISTABILITY
  • GEMINI
  • FIREFLY
  • FLUX.1
  • CANVA
  • IDEOGRAM
Private · Deleted the moment analysis ends

How every scan runs

Read the method

  1. Provenance

    C2PA content credentials and EXIF metadata, checked first

  2. Watermark & Pixels

    OpenAI and Google AI watermarks read, then a detection model scores the image

  3. Evidence Chain

    Every finding itemised and graded, not reduced to a yes or no

  • Files deleted after analysis
  • No account required

How to detect a DALL-E image

Most DALL-E images in circulation were never made on OpenAI's own site. They came out of ChatGPT, Bing Image Creator or Microsoft Copilot - products used by hundreds of millions of people who may not realize they are using DALL-E at all. Detecting DALL-E images involves a few distinct checks. The first is C2PA provenance verification: DALL-E 3 embeds cryptographically signed Content Credentials into every output image using the C2PA (Coalition for Content Provenance and Authenticity) standard. When this metadata is intact, FauxLens reads it, and a signed AI-tool credential can end the scan on its own. Images made in ChatGPT since May 2026 also carry an invisible OpenAI watermark, which FauxLens reads and reports as OpenAI rather than as a specific model. The last is pixel analysis by a specialist detection model, which carries the scan when C2PA has been stripped. Stripping happens whenever an image is screenshotted, re-saved at a different quality, shared via a messaging app that re-encodes images, or processed by any tool that does not preserve XMP metadata. In this case, the detection model scores the pixels and says whether the image looks AI-generated, partly changed by AI or real. It works on DALL-E 2 and DALL-E 3 outputs from every access point, though heavy compression lowers confidence.

C2PA Content Credentials: DALL-E's Built-In Detection Signal

C2PA (Coalition for Content Provenance and Authenticity) is an open technical standard developed jointly by Adobe, Microsoft, Google, the BBC, and others to create a verifiable chain of custody for digital content. OpenAI implemented C2PA in DALL-E 3 and all subsequent versions. Every image generated through ChatGPT's image generation feature, the DALL-E API, or DALL-E via Microsoft Copilot automatically receives a cryptographically signed Content Credentials payload embedded in the file's XMP metadata. This payload contains the assertion that the content was AI-generated, a timestamp of generation, the issuer identifier (OpenAI), and a cryptographic signature that verifies the payload has not been tampered with. FauxLens reads and verifies this signature as its primary detection signal. When C2PA is present and the signature validates, the result is as definitive as a certificate of authenticity. When C2PA is absent (because the image was screenshotted, shared through a platform that strips metadata, or saved with a tool that does not preserve XMP), our forensic pipeline takes over. Metadata absence is not evidence of authenticity: it only means the provenance signal is unavailable. After metadata stripping, FauxLens still has the OpenAI watermark on newer ChatGPT images and a detection model that scores the pixels, though heavy compression lowers the confidence of the pixel score.

What changed between DALL-E 2 and 3

DALL-E 2 (released 2022) used a CLIP-guided diffusion architecture that operated in pixel space rather than latent space, producing images with more visible generation artifacts - characteristic blocky patterns in flat color regions at higher frequencies and a distinctive noise profile. DALL-E 2 does not embed C2PA metadata, so detection relies entirely on pixel-level forensics. Detection accuracy for DALL-E 2 approaches 97% because the artifacts are consistent and strong. DALL-E 3 (released 2023) switched to a latent diffusion architecture with significantly improved image quality and text rendering. It introduced C2PA metadata embedding as the primary detection signal. Its output looks noticeably cleaner than DALL-E 2, with fewer of the blocky patches that gave the older model away. DALL-E 3 images that have had metadata stripped can still be caught by the detection model, but are slightly more challenging to detect than DALL-E 2 because the overall image quality is higher. DALL-E 3 also generates images with coherent text, a capability DALL-E 2 largely lacked, which removed garbled lettering as an easy visual clue.

Where DALL-E images come from, and how they differ from photographs

Most people who encounter DALL-E images in 2026 do not know they are from DALL-E, because the model is embedded in multiple consumer products under different names. ChatGPT's image generation feature uses DALL-E 3 and newer OpenAI image models - all images generated through the ChatGPT interface carry C2PA Content Credentials. Bing Image Creator, Microsoft's consumer image generation tool, is powered by DALL-E and has generated an enormous volume of images since its 2023 launch. Microsoft Copilot's image generation capability also runs on DALL-E infrastructure. These products share the same underlying DALL-E generation pipeline and produce images with the same forensic signatures. The practical implication for detection: an image shared as something found online may have been generated by someone using ChatGPT casually, without any intent to deceive, and without the sharer knowing it was AI-generated. The C2PA metadata, if intact, will identify it definitively. Without it, the detection model can still judge the pixels, though copies shared through Instagram, WhatsApp, or similar apps are harder because they are heavily compressed.

Read more

DALL-E images can be told apart from real photographs in more than one way. The strongest is provenance: a C2PA credential naming OpenAI or, on images made in ChatGPT since May 2026, the invisible OpenAI watermark, which FauxLens reports as OpenAI rather than naming a model. Metadata can also clear a real photo, since a camera record is evidence the picture was taken rather than generated. A file with no camera fields (make, model, lens, ISO, shutter speed, GPS) proves nothing, because screenshots and messaging apps strip them from real photos too, so FauxLens never treats that absence as evidence. When provenance is gone, a specialist detection model scores the pixels. You can also look for yourself: shadows in DALL-E images are generally more internally consistent than in Midjourney outputs, but small errors in shadows and reflections still show up on a close look.

Is that photo or video real?

Frequently asked questions

FauxLens detects DALL-E 3 images through both C2PA provenance metadata verification and forensic pixel analysis. When C2PA metadata is intact, detection is definitive. When C2PA has been stripped by screenshotting or re-saving, FauxLens checks for the invisible OpenAI watermark and has a detection model score the pixels.

DALL-E 3 and later versions embed C2PA Content Credentials - cryptographically signed metadata that identifies the image as AI-generated by OpenAI. DALL-E 2 does not embed C2PA. FauxLens checks for C2PA data as its primary detection signal, then falls back to forensic pixel analysis when metadata is absent or has been stripped.

Yes. ChatGPT's image generation feature uses DALL-E 3 and later OpenAI models. All images generated through ChatGPT automatically carry C2PA Content Credentials that FauxLens detects. Images made in ChatGPT since May 2026 also carry an invisible OpenAI watermark that FauxLens reads, and if both are missing, a detection model still scores the pixels.

Screenshotting a DALL-E image creates a new file with no XMP or EXIF metadata from the original - C2PA credentials are lost. FauxLens then checks for the invisible OpenAI watermark carried by ChatGPT images since May 2026, and has a specialist detection model score the pixels. A screenshot of an AI image can still be caught this way, though confidence drops on small or heavily re-compressed copies.

OpenAI has continued developing its image generation capabilities beyond DALL-E 3, with updates embedded in ChatGPT's image generation feature. FauxLens reads the invisible OpenAI watermark that every ChatGPT image has carried since May 2026, and reports it as OpenAI rather than naming a model. All versions embed C2PA metadata when accessed through official OpenAI products, making detection straightforward when metadata is intact.

C2PA embeds a signed manifest in the image file using XMP metadata. The manifest contains assertions about the content, such as that it was AI-generated, plus a cryptographic signature created with a private key held by the issuer (OpenAI). FauxLens reads the manifest, verifies the signature against the issuer's public certificate, and returns the result. A valid signature proves the assertion is authentic and untampered. The signature covers the image content, so even a one-pixel change invalidates it.

Yes. DALL-E supports in-canvas editing - erasing regions of a real photo and having DALL-E fill them in (inpainting). The detection model tells a real photo that AI has changed apart from an image AI made from scratch: the scan comes back as likely AI-generated, with an evidence line noting that an AI tool altered only part of the image. It does not mark which region was changed, and a small edit is harder to catch than a large one.

OpenAI does not embed visible watermarks in DALL-E images. Their approach to provenance is C2PA Content Credentials - invisible cryptographic metadata rather than visible branding. This means DALL-E images look visually clean but carry machine-readable provenance that FauxLens can read and verify when the metadata has not been stripped.

Keep exploring

Guides

More tools

Detect Flux AI Images Detect Midjourney AI Images Detect Stable Diffusion Images Free AI Image Detector: Was This Photo Made by AI? Free Deepfake Detector Free Fake Image Detector