Detect Flux AI Images
Upload an image and find out whether Black Forest Labs' Flux made it. FauxLens covers Flux Dev, Flux Pro and Flux Schnell, including fine-tuned community models, and answers in seconds.
Upload the image or video. It is deleted after the check.Drop a photo or video here, or click to upload. It is deleted after the check.
JPG PNG HEIC · MP4
Detects content from
- OPENAI
- MIDJOURNEY
- STABILITY
- GEMINI
- FIREFLY
- FLUX.1
- CANVA
- IDEOGRAM
How every scan runs
Read the methodProvenance
C2PA content credentials and EXIF metadata, checked first
Watermark & Pixels
OpenAI and Google AI watermarks read, then a detection model scores the image
Evidence Chain
Every finding itemised and graded, not reduced to a yes or no
- Files deleted after analysis
- No account required
What Flux.1 is, and why it is hard
Flux.1 is a family of AI image generators developed by Black Forest Labs, a company founded by former Stability AI researchers including Robin Rombach, the lead author of the original Stable Diffusion paper. Released in 2024, Flux.1 achieved state-of-the-art image quality that surpassed Midjourney v6 and SDXL on multiple benchmark evaluations: particularly for photorealistic portraits, coherent text rendering, and complex compositional scenes. Flux.1 is architecturally distinct from previous diffusion models. Traditional diffusion models like Stable Diffusion and Midjourney use score-based stochastic differential equations to progressively denoise images. Flux uses rectified flow matching, a different mathematical framework that defines straighter trajectories from noise to image in the sampling process. This produces cleaner, less noisy images with fewer of the visible artifacts that made earlier SD outputs easier to detect. The dual-stream transformer architecture (DiT - Diffusion Transformer) processes image patches and text conditioning in parallel streams before merging, which produces a different pattern of spatial attention artifacts than the U-Net architectures used by SD 1.5 and SDXL. Flux.1 Dev is open-source with 12 billion parameters - significantly larger than SDXL's 6.6B, and runs locally without any API or moderation layer, making it the preferred tool for high-quality fraud imagery in 2025 and 2026.
The fingerprints, and how they differ across Dev, Pro and Schnell
Flux images are cleaner than most diffusion output, and that cleanliness is itself a signature. The rectified flow sampling process follows a straighter path from noise to image than score-based diffusion, and the result tends to look unusually clean: smooth gradients, tidy skin, and backgrounds with little of the grain a real camera records. Flux also renders readable text, so garbled signs are no longer a reliable clue. Because Flux carries no invisible watermark from OpenAI or Google, FauxLens leans on two things for it: any generator signature left in the file metadata, and a specialist AI-image detection model that scores the pixels. That model judges whether AI made the image rather than naming Flux, and it still works through typical post-processing such as JPEG compression and moderate resizing, with lower confidence on heavily compressed or small copies.
Read moreShow less
The three main Flux.1 variants differ in sampling steps, architecture details, and access method, which produces outputs that look slightly different from each other. Flux.1 Dev is the open-source, full-quality variant with 12B parameters. It uses the full rectified flow sampling schedule (typically 20-50 steps) and produces the highest quality outputs of the three. Detection accuracy on Flux Dev outputs exceeds 91%. Flux.1 Schnell is a distilled variant trained to produce acceptable quality in 1-4 sampling steps - significantly fewer than Dev. The distillation process compresses the sampling trajectory, which changes the noise characteristics of the output. Schnell images show more compression-like artifacts than Dev outputs because the distilled model has fewer steps to refine detail. These artifacts make Schnell images slightly easier to detect than Dev outputs, despite Schnell's lower overall image quality. Flux.1 Pro is an API-only commercial variant that Black Forest Labs periodically updates with undisclosed architectural refinements. Pro outputs look slightly different from Dev outputs, reflecting these refinements. FauxLens does not tell the three variants apart; its detection model judges whether AI made the image, whichever variant was used. Community fine-tunes based on Flux Dev (available on Civitai and HuggingFace) inherit the Dev forensic fingerprint while adding style-specific artifacts.
Where Flux images have shown up
Flux.1's combination of open-source availability, high image quality, and local execution without moderation has made it the fastest-growing tool for AI-generated fraud imagery since its 2024 release. In romance scams, Flux has largely replaced Midjourney for operators who want consistent high-quality persona images without Discord attribution - Flux runs locally with no service trail. Romance scam imagery became noticeably more sophisticated after Flux.1's public release. In synthetic identity fraud, where AI-generated faces are used to create fraudulent identity documents, Flux's photorealistic portrait quality has made it the preferred tool. The document fraud pipeline typically involves generating a Flux portrait, then compositing it into a genuine document template. In political disinformation, Flux-generated imagery appeared in documented influence operations targeting the 2025 UK local elections and several European municipal elections, where photorealistic fabricated imagery of politicians was distributed through Telegram channels before fact-checkers identified it as AI-generated. Flux's lack of any visible watermark or platform attribution, unlike Midjourney's subtle Discord-linked metadata, makes attribution of Flux-generated disinformation significantly harder without forensic analysis tools.
Keeping up as the models move
Keeping Flux detection current is a continuous engineering challenge because Black Forest Labs actively develops the model and the community produces new fine-tunes weekly. FauxLens does not rely on recognizing Flux itself. Its pixel analysis comes from a specialist third-party AI-image detection model that judges whether AI made an image rather than which model did, so a new fine-tune is not automatically a blind spot, though a brand-new generator can lower accuracy until detectors catch up. That model is paired with checks that do not depend on the pixels at all: C2PA content credentials, EXIF and XMP metadata, and the invisible watermark OpenAI and Google embed in images their own tools make. If the detection model is unavailable, a multimodal vision model takes its place. The result is a verdict with a confidence score and evidence lines, so you can see how much of the answer rests on the file itself and how much on the pixels. New releases such as the Flux 1.1 Pro update in late 2024 are a reminder that no detector stays current on its own.
Is that photo or video real?
Frequently asked questions
FauxLens detects images from all Flux.1 variants (Dev, Pro, and Schnell) as well as community fine-tunes based on Flux Dev. A specialist detection model scores the pixels and judges whether AI made the image, and any generator signature left in the metadata is read too.
Both leave distinct mathematical fingerprints, but Flux.1 Dev is marginally more challenging to detect than Midjourney v5 because the rectified flow architecture produces cleaner images with less visible noise. FauxLens detects both with comparable overall accuracy, judging whether AI made the image rather than which generator did.
Flux.1 Dev is open source under a non-commercial license. Flux.1 Schnell is released under the Apache 2.0 license, permitting commercial use. Flux.1 Pro is a closed API-only product. The open-source availability of Dev and Schnell means Flux can be run locally without any API, content moderation, or usage tracking - which is why it has become popular for fraud use cases.
Yes, the forensic fingerprints are embedded by the Flux model itself during generation, not by the serving infrastructure. There is no difference between images generated by Flux running locally versus through an API. Running Flux locally does not affect the mathematical artifacts in the output.
In independent benchmark evaluations from 2024-2025, Flux.1 Dev produces photorealistic portrait quality comparable to or exceeding DALL-E 3, particularly for images of people, architecture, and product photography. The two models have different artifact profiles (DALL-E 3 carries C2PA metadata, while Flux carries no provenance metadata), making Flux outputs harder to attribute through metadata verification and requiring full forensic pixel analysis.
Flux.1 does not embed any visible watermarks or invisible cryptographic watermarks in generated images. Unlike DALL-E 3 (which uses C2PA Content Credentials), Flux has no provenance metadata system. For Flux images FauxLens relies mostly on its specialist detection model, which scores the pixels, plus any generator signature left in the file metadata.
Yes, with slightly reduced confidence. Social media platforms (Instagram, Twitter/X, WhatsApp) aggressively re-compress uploaded images, which degrades some forensic signals. The detection model still scores what is left, but heavily compressed or small copies are where it is weakest. Detection accuracy drops by approximately 5-10% on heavily compressed social media versions versus original PNG files.
Rectified flow is a generative modeling framework that learns to map noise distributions to image distributions along straighter trajectories than traditional score-based diffusion models. In practice, this means Flux reaches its final image state in fewer, larger steps than DDPM-based models. The straighter sampling trajectory produces different intermediate noise states, which leaves a different residual noise signature in the final image compared to SD or Midjourney. FauxLens does not use this to single out Flux; its detection model judges whether an image is AI-generated rather than which generator made it.
FauxLens achieves over 91% detection accuracy on Flux.1 Dev outputs, approximately 93% on Flux.1 Schnell (which shows stronger compression-like artifacts due to distillation), and approximately 90% on Flux.1 Pro. These figures are for images with minimal post-processing. Heavily post-processed images or images run through additional AI upscalers show reduced confidence.
Flux can be used for face replacement via inpainting - erasing a face region and using Flux to generate a replacement. That is a real photo with part of it generated. It comes back as likely AI-generated, with an evidence line noting that an AI tool altered only part of the image, and FauxLens does not mark which region changed. Dedicated face-swap tools (DeepFaceLab, FaceSwap) are architecturally different from Flux and leave different signatures.
Keep exploring
Guides
- The Science of Deception: How AI Detection WorksA detailed guide to the digital forensics of AI image detection, from Error Level Analysis (ELA) and JPEG compression artifacts to Photo Response Non-Uniformity (PRNU), and which checks a FauxLens scan actually runs.
- Every AI Image Has a Hidden Fingerprint - Here's How Forensics Finds ItEvery AI model leaves an invisible fingerprint in the images it generates. GAN fingerprints are the forensic equivalent of ballistic markings-unique to the weapon, invisible to the eye, detectable by science.
More tools
Detect while browsing: try the Chrome extension
Right-click any image · No account required