The Fake Mia Goth Selfie: How to Tell If an Image Is AI

The moment

What happened with the fake Mia Goth selfie?

An Instagram account with the handle @miagoth_ posted an AI-altered selfie of the actor wearing a T-shirt with a pointed message about an ex. It went up days after Goth’s divorce filing made headlines, so it looked like a reply. According to Variety (October 13, 2026), several outlets reported it as genuine. Rolling Stone reported that People published a story treating the photo as Goth’s own post, and that story has since been removed.

Goth does not use social media and has never had a public Instagram account. Spokesperson Christine Tripicchio told The Verge the team had filed multiple complaints about the account posting fake images. Tripicchio called the account’s blue check “disconcerting on a fundamental level.” A Meta spokesperson told Variety the account was removed for impersonating a public figure.

Why it worked

Why did so many people believe it?

The fake worked because of everything around the pixels:

• The handle looked official. @miagoth_ is one underscore away from a name people expect.
• The badge looked official. The account had about a million followers and a blue check, even though its bio said fan account. As Rolling Stone noted, paid Meta Verified badges look nearly identical to standard ones.
• The timing fit the story. A photo that matches the headline of the week feels true before anyone checks it.
• Rules allow it. Variety reports that Meta does not ban AI images of real people. It labels AI-modified content instead, and a label only helps if it is applied and noticed.

Your takeaway

How to tell if an image is AI: a 5-step check

Pause before you share. This takes about two minutes:

1. Find the first poster. Who shared it first? Open the profile. Read the bio for words like fan, parody or AI.
2. Confirm the account is theirs. Look for a link from the person’s official website, agency or another account you already trust. A badge alone is not proof.
3. Look for confirmation. Has a trusted outlet or the person’s team confirmed it? If only reposts and screenshots exist, wait.
4. Run it through a checker. Upload the original file to a free tool like SynthID Detector (below). Remember that “no watermark found” does not mean real.
5. Then look closely. Text on clothing and signs, jewelry, ears, hands, reflections, and skin that looks airbrushed everywhere at once. Treat these as clues, not a verdict.

I found an image online and I want to check it before I share it.
Where I saw it: [platform, account name, account bio]
What it claims to show: [one sentence]
When it was posted: [date]
Act as a careful fact-checker. Don't tell me if it's real or fake.
Instead, give me a short checklist of exactly what to verify:
1. Questions to ask about the account that posted it
2. Where to look for the original source or an official statement
3. Which visual details to inspect in this kind of image
4. What would count as solid confirmation either way
Keep it under 150 words.

You stay the investigator; AI only hands you the checklist. For moving images, our guide on how to tell if a video call is AI covers live deepfakes and verification questions.

The new tool

Is there a free AI image checker you can trust?

There is a new one, with honest limits. On October 7, 2026, Google opened SynthID Detector to everyone. You upload an image, video or audio file, and it looks for SynthID, an invisible watermark. It now reads content from Google plus partners OpenAI, NVIDIA and Kakao, with Apple listed as coming soon. Google says it has watermarked over 180 billion images and videos.

The catch: it only finds watermarks it knows. As Innovation Village reported, it is free with a Google, Apple or OpenAI sign-in, watermarks can sometimes be stripped, and it cannot separate a fully AI-made image from a real photo lightly edited with AI. A clean result is one data point. The five-step check is still the whole picture.

The debate

Should platforms ban AI images of real people?

The case for stricter rules. Goth’s team argued Instagram put the spread of forged images ahead of protecting a person’s likeness, after filing multiple complaints while the fake images stayed up for days. Variety notes that Scarlett Johansson, Cate Blanchett and Steven Soderbergh have also spoken out against deepfakes, and that Blanchett cofounded the Human Consent Foundation so people can register whether AI models may use their likeness.

The case for labels instead of bans. Meta’s current approach labels AI content rather than removing it, and the account was taken down once it was judged to be impersonation. People who favor labels point out that parody, fan art and commentary have long used real faces, and that a blanket ban would sweep in creative work along with fakes.

What this case does show: badges and labels only help when people can read them at a glance. Pixel Pinkscape’s line is simple. Real faces need consent, and AI content gets a clear label. Our guide on whether you have to disclose AI content explains how creators do that well.

Free resources

Free resources for spotting AI images

• SynthID Detector — Google’s free checker for watermarked images, video and audio from Google, OpenAI, NVIDIA and Kakao. Sign-in required.
• Checkology by the News Literacy Project — free news literacy lessons for educators, including a lesson on algorithms and generative AI.
• FinalFile by PageProof — a free web tool that checks whether a file carries Content Credentials showing its origin and any AI involvement.
• The Pixel Pinkscape Dreamers community — a free Facebook group where you can ask real people, “Does this look real to you?”

FAQ

Frequently asked questions