Google’s deepfake detector system used to debunk McConnell hoax pic

What actually happened with the McConnell deepfake?

A hoax image showing US Senator Mitch McConnell hospitalised went viral on 8 July 2026 before Snopes debunked it using Google's SynthID watermark, which was embedded invisibly in the picture's pixels and confirmed the image was AI generated.

SynthID works differently to older watermarking methods. It embeds its signature directly into the pixel data itself, not in file metadata, which means it survives screenshots, resizing and compression as the image gets shared and reshared across platforms. That is precisely what let Snopes catch it even after the picture had circulated.

Why should a UK small business owner care about this?

Because the same technology that debunked a political hoax is now sitting in tools you already use, and because impersonation fraud in the UK is rising fast enough that ignoring this is a real business risk, not just a curiosity.

UK payment fraud losses hit £1.28 billion in 2025, with authorised push payment fraud, the type that relies heavily on impersonation, up 19% to £576.4 million. Separately, the Sumsub Identity Fraud Report 2025 recorded a 94% rise in UK deepfake attempts, and a 2025 Gartner survey found more than six in ten organisations had already faced a deepfake attack.

If you run a business with a public face online, whether that is your team, your reviews, your before and after photos or your owner's face on marketing, you are now a plausible target for fake images built to damage trust or impersonate your brand.

Could someone fake a photo of my business or staff?

Yes, and it is easier than most owners assume. Anyone with access to a modern AI image generator can produce a convincing fake review photo, a fabricated "incident" at your premises, or a doctored screenshot designed to look like a complaint or scandal.

The good news is that if the fake was made using Gemini, which has carried SynthID since Google I/O 2025, or another participating tool (OpenAI joined the same watermarking effort in May 2026), it can potentially be checked and disproved.

How do I actually check if an image is AI generated?

Use Google's public SynthID Detector portal to upload the image, audio clip, video or piece of text directly. Detection is also built into the Gemini app and into "About this image" inside Search and Chrome, so you do not always need a separate tool.

1. If a suspicious image appears online (a fake review, a fabricated incident, an impersonation of your brand), save it and run it through the SynthID Detector or "About this image" first.
2. If it flags as AI generated with SynthID, you have concrete evidence to report to the platform, Google, or trading standards.
3. If it does not flag, that does not prove it is real. It may have been made with a tool that has not adopted SynthID.

What's the catch with SynthID?

It only detects images made with AI tools that have actually adopted the watermark. Anything generated on a non participating tool will not be caught, so SynthID is a useful check, not a guarantee.

This is the point most coverage of this story misses. A single detector, however good, cannot replace basic verification habits. Treat SynthID as one layer, alongside reverse image searches, checking the original source, and simply calling the person or business directly before reacting to anything alarming online.

Braynex Services' take: this is another reason to own your presence, not rent it

We have said for a while that renting your online presence, a Facebook page, a Linktree, a directory listing, is a liability because you do not control the terms, the data, or what gets posted alongside your name. This story adds a new dimension: you also do not control what gets faked about you, and rented platforms are slower and less accountable when it happens.

A business with its own website and a verified Google Business Profile has a clearer, more authoritative home for the truth. When we helped a Leeds salon verify and properly optimise its Google Business Profile, it went from zero online presence to over 40 monthly calls within three weeks, purely from Google Maps. That same verified profile is exactly the kind of trusted, owned asset that gives customers and journalists a fast way to check what is real, rather than relying on whatever is circulating on social media.

Our practical advice: bookmark the SynthID Detector, teach whoever manages your social media and reviews to use it, and treat it as one part of a wider habit of owning your online presence rather than depending on platforms you do not control.

If you are not sure how exposed your business is online, whether that is fake reviews, an outdated Google listing, or a presence you do not fully control, Braynex Services offers a free audit at braynexservices.com. We will show you exactly where you stand and what to fix first.