Can Text Messages Be Faked? What to Know About AI-Generated Fake Screenshots
Yes — and it doesn't take special skill or software. Free websites that generate a fake iMessage or WhatsApp screenshot, with any text you type in, have existed for years. That's not a reason for panic about every screenshot you'll ever see. It is a good reason to understand how these tools actually work, what a screenshot can and can't prove on its own, and why that gap matters more now than it used to.
How fake screenshot generators actually work
A number of free websites and apps let anyone recreate the look of a text conversation: pick a platform (iMessage's blue bubbles, Android's green ones, WhatsApp, Instagram DMs), type both sides of the exchange, set a contact name and photo, adjust the timestamp and battery icon, and download the result as an image. Many require no account and no payment. This isn't a new, exotic capability — these mockup generators have been used for years for pranks, memes, film props, and app design previews, and most of them say exactly that in their own terms of use. What's changed is that some now use AI to write the fake dialogue for you from a short description, and the visual fidelity of the output has gotten good enough that, at a glance, it's hard to tell from a real screenshot. The underlying trick — a picture built to look like a screen, containing whatever text someone typed into it — predates generative AI by a decade or more; AI mainly made the fake text itself faster to produce.
What a screenshot fundamentally can't prove
A screenshot is an image — a picture of pixels. On its own, it doesn't carry the data that actually lives behind a real conversation on a phone: system-written timestamps, sender and recipient identifiers, delivery and read receipts, and the structural record a messaging app keeps in its own database. None of that travels with a PNG. That's true whether the screenshot is completely fabricated, or a real conversation that's been cropped, edited, or had a message added or removed — both are largely indistinguishable from a genuine screenshot to an ordinary viewer. A screenshot shows what an image displays. It doesn't, by itself, show whether the underlying conversation happened at all.
A real case where fabricated texts got caught
In Rossbach v. Montefiore Medical Center, a federal court found that the key evidence behind a plaintiff's harassment claims — a photo purporting to show sexually explicit text messages — had been fabricated. The plaintiff testified she'd taken the photo on an older iPhone model because her original phone's screen was too cracked to screenshot. Forensic review found that the image rendered an emoji design that Apple hadn't introduced until an iOS version her stated phone couldn't even run. The district court dismissed her claims with prejudice and imposed sanctions; the Second Circuit affirmed in 2023. It's a useful real example, and also an honest one to sit with: what caught the fabrication wasn't something visible to an ordinary reader. It took a forensic-level detail — which emoji design shipped with which iOS version, cross-referenced against a specific phone model — that almost nobody would think to check.
What a native export gives you that a screenshot can't
A native export — the conversation pulled directly from the platform, not a photo or screen capture of it — keeps the timestamps, sender information, and full message sequence intact, written by the platform itself rather than recreated by hand. That's a structurally different kind of record than an image, and it's the reason a native export tends to be treated as stronger evidence. We go into this in more detail in Why Screenshots Alone Aren't Enough for Court.
If you suspect a screenshot has been faked
Being honest here matters more than being reassuring: there's no simple, universal way for an ordinary person to look at an image and know whether it's fake. The Rossbach case above is a reminder of that — the tell was a forensic detail, not something visible on the screen. If authenticity of a piece of evidence is genuinely in dispute, that's what the evidentiary authentication process and digital forensics examiners exist for, not a trick you can perform yourself. Our overview of how electronic evidence gets authenticated under FRE 901 and 902 covers that process in more detail. What you can control is your own evidence: producing a native export instead of a screenshot removes the question entirely, because there's no re-creation step for anyone to have tampered with.
This is a narrower, more current version of a broader trend we cover in Deepfakes, AI, and the Future of Text Message Evidence — that post looks at a 2026 court ruling and a proposed federal evidence rule, both about how courts are responding to AI-generated and machine-generated evidence more broadly.
Verity is a forensic and behavioral analysis tool, not a law firm and not a digital forensics lab. Nothing here is a claim that Verity can determine whether a specific screenshot is authentic or fabricated. This is general information, not legal advice — if authenticity of evidence is in dispute in your case, talk to an attorney about your options, including a forensic examiner.
Work from a native export, not a screenshot
Upload your export and get a free health score in minutes — the analysis runs in your browser, nothing is uploaded.
Get your free health score →