Skip to content
AI & Technology🔥 Trending

How to spot AI-generated images with Google Lens & FotoForensics

Learn how to spot AI-generated images using Google Lens, FotoForensics, Yandex and simple visual clues. Quick steps you can use on Android and iPhone.

Smartphone showing image analysis with Lens and FotoForensics to spot AI-generated images
Quick answer

Look for visual tells (hands, text, shadows, repeating patterns), run a reverse image search with Google Lens or Yandex, and check file analysis with FotoForensics (ELA) and metadata. Combine a quick eyeball check with one or two free tools — that gives the fastest, most practical result on WhatsApp or social apps.

Key takeaways

  • Start with quick visual checks: hands, text, shadows and background details.
  • Use Google Lens (Android/iPhone) or Yandex for reverse image search of photos shared on WhatsApp.
  • Use FotoForensics’ Error Level Analysis and metadata view to look for edits or missing EXIF.
  • No single test is foolproof — combine visual clues, reverse search and metadata/ELA.
  • If it matters (legal, safety), preserve the original file and consult professionals.
In this article
  1. Who this is for and what you need
  2. Step-by-step
  3. Common problems and fixes
  4. Tips and best practices
  5. Quick comparison table of tools
  6. Frequently asked questions
  7. Sources

Who this is for and what you need

This guide is for anyone in India who gets suspicious images on WhatsApp, Instagram or Facebook and wants a fast, no-technical way to check if they are AI-generated. You don’t need to be a tech expert — just a smartphone and free web tools.

What you need: a working Android or iPhone with the image saved or accessible, and an internet connection. We use three free checks: quick visual clues, Google Lens or Yandex reverse search, and FotoForensics (Error Level Analysis and metadata).

Step-by-step

  1. Save the original image — Save the image from WhatsApp or the social app before you forward or edit it. This preserves file metadata and the exact pixels for analysis.
  2. Do a 30-second visual check — Look for obvious tells: wrong number of fingers, jagged or garbled text on signs, strangely smooth skin, mismatched shadows or multiple light directions, repeating background elements, or odd reflections in eyes. If several tells appear, be more suspicious.
  3. Run Google Lens reverse image search — On Android: open the image in Google Photos or Chrome, tap Lens (or long-press image in Chrome > Search image with Google). On iPhone: install Google app or Google Photos, open the image and tap Lens. Look for earlier versions of the image, similar images, or a source website. If Lens finds a stock photo or the same image on a news site dated earlier, that’s a clue.
  4. Try Yandex image search — Yandex sometimes finds different copies that Google misses. On desktop, drag the image to Yandex Images. On phone, open the image in your browser and use the Yandex Images search-by-image option. If Yandex finds multiple near-identical but unrelated results, it could be AI or heavily edited content.
  5. Check for obvious labels or watermarks — Some AI tools add visible watermarks or stamps (or leave invisible signals like SynthID that some Google tools detect). Visible watermarks are straightforward; absence of watermark is not proof the image is real because watermarks can be removed.
  6. Upload to FotoForensics for Error Level Analysis (ELA) — On a desktop or phone browser, go to FotoForensics, upload the saved image and view the ELA result. ELA highlights areas with different compression levels. Large, contrasting patches or wildly different ELA patterns can indicate editing or compositing. Uniform ELA usually suggests the image is unedited, but uniformity isn’t a guarantee of authenticity.
  7. Check metadata / EXIF in FotoForensics — Look at the metadata panel. Real photos often show camera make/model, lens and timestamps. Many AI-generated images or screenshots have missing EXIF or generic export tags. Note: social apps often strip EXIF, so missing metadata on WhatsApp images is common and not definitive.
  8. Compare results and decide — If visual tells, reverse search, and ELA all point to manipulation or no real source, the image is likely AI-generated or edited. If results conflict (for example, clear source found but ELA shows possible edits), treat the image as suspicious and seek the original uploader or source for confirmation.
  9. Preserve evidence if it matters — If the image is part of harassment, scams or possible crime, keep the original file, screenshots of your checks, and the chat where it was shared. For serious cases, contact local authorities or a verified fact-checker service.

Android vs iPhone notes

  • Google Lens: Built into many Android phones (Photos, Chrome). On iPhone use the Google app or Google Photos to access Lens.
  • FotoForensics: Works in any mobile browser or desktop. For best results, upload the saved original image rather than a screenshot.
  • EXIF: iPhone Photos may show richer metadata, but iMessage/WhatsApp often strip it. Always check the saved file, not a forwarded screenshot.

Also read: 7-step guide to Google Gemini Memory: view, edit, delete

Common problems and fixes

  • Problem: WhatsApp image has no metadata — Fix: Ask the sender to forward the original file (use WhatsApp’s Document option to send the full image) or request the original camera file. Screenshots and forwards usually strip EXIF.
  • Problem: Lens or Yandex finds no matches — Fix: That can mean the image is new, private, or AI-generated. Combine with visual checks and FotoForensics. A single negative reverse-search result is not proof.
  • Problem: FotoForensics ELA is confusing — Fix: ELA is a hint, not a verdict. Large, bright blocks in ELA suggest editing; small differences may be compression artefacts. If uncertain, ask a trained fact-checker or use a second detector.
  • Problem: The image has readable text but it’s garbled — Fix: Garbled or nonsensical text in signs, books or labels is a strong AI clue. Zoom in and read every word. If letters don’t form real words, treat the image with suspicion.

Tips and best practices

  • Combine signals: visual clues + reverse search + ELA/metadata — more than one match increases confidence.
  • If an image looks too perfect (flawless skin, perfect symmetry), be suspicious — many generators optimise for beauty.
  • Save images immediately if you may need proof later. Use the Document share option in WhatsApp to keep the original file intact.
  • For newsworthy or legal matters, rely on professional fact-checkers or forensic labs; consumer tools are good for quick checks only.
  • Teach family and elders simple checks: look at hands, read text in the picture, and run a quick Lens search before forwarding.

Quick comparison table of tools

Tool Best for Limitations
Google Lens Fast reverse search, context, SynthID detection in some Google tools May miss private or new images; not a proof
Yandex Images Alternate reverse search that finds different matches Interface less familiar to some users
FotoForensics (ELA) Highlights edits and shows metadata ELA is interpretive and social apps often strip EXIF

Use these steps whenever you see an image that could mislead people. Simple, repeatable checks work best for most WhatsApp and social-app cases.

ViralDuniya verdict

For everyday checks on WhatsApp and social apps, combine a quick visual inspection with Google Lens/Yandex reverse search and a FotoForensics ELA/metadata check. If the image matters legally or for safety, preserve the original and consult professional fact-checkers—consumer tools are helpful but not definitive.

Frequently asked questions

Can Google Lens always tell if an image is AI-generated?

No. Google Lens helps find earlier copies and may surface labels like SynthID in some Google products, but it does not always identify AI-made images. Use Lens as a fast first step, then confirm with other checks.

Does FotoForensics prove an image is AI-made?

No. FotoForensics’ ELA highlights areas with different compression, which can indicate edits. It does not prove an image was generated by AI. Treat ELA as one clue alongside visual checks and reverse searches.

Why do WhatsApp images lose metadata?

WhatsApp compresses and re-encodes images to save data, which removes EXIF metadata. To keep the original file, ask the sender to use WhatsApp’s Document sharing option or a cloud link.

What visual signs should I look at first?

Look at hands and fingers, readable text (signs, labels), reflections/eye highlights, shadow direction, and repeating or blurry background people or objects. These are common AI weak spots.

Is a reverse image search enough to verify an image?

Often it helps but is not enough alone. Reverse search can show original sources or earlier instances, which is powerful. If it finds nothing, combine that with ELA and visual clues before concluding.

Sources

  1. Fact check: How do I spot AI-generated images? – DW.com

Ask a question or share your experience

Your email address will not be published. Required fields are marked *