How To Verify Screenshots Are Not AI Generated

Fake screenshots of social posts, text messages and news headlines were common long before AI. Anyone can edit a web page in their browser and screenshot it. AI makes it even easier to generate convincing chat logs or posts. Either way, the check is the same: find the original.

Find the original

• Social posts: search the account directly for the post, or search the exact wording in quotes.
• Deleted posts: check the Wayback Machine and archive.today, and search for news coverage of the post.
• News headlines: visit the outlet's site and search for the article.
• Messages and emails: ask for the original or for someone else in the conversation to confirm it.

Clues of editing or generation

• Fonts and spacing that differ from the real app or site.
• Layout details that don't match the app's current design, such as old icons or wrong button positions.
• Numbers that don't add up: likes, timestamps or dates that are impossible or inconsistent.
• Profile pictures and names that don't match the real account.
• Garbled small text in AI-generated screenshots, such as in menus or timestamps.

Why file checks don't help much

Screenshots don't contain information about what was on screen. Their metadata, if any, shows the device or screenshot tool. A metadata reader such as AI Image Detector (made by Patrick Bushe, who runs this site) will usually show "No Metadata Found" or only screenshot software. That can't confirm the content is real.

Check the context

Who first shared the screenshot, and why? Screenshots that conveniently prove an outrageous claim, with no link to the original, deserve the most suspicion. Reputable fact-checkers often investigate viral screenshots; search for the claim with "fact check."

When you can't verify it

Don't share it as fact. If you discuss it, say it couldn't be verified. See checking viral posts.