What happens when the map becomes a liar?
You’ve probably used Google Maps to direct a customer to your shop, or dropped a pin to show a supplier where your warehouse sits. Google Earth is the view you pull up when you want to check a location from above before you commit to it. For years, that satellite view was the closest thing to objective truth you could get without visiting the place yourself.
Last Thursday, Google turned that truth into a suggestion. A new feature in Google Earth let anyone type a text prompt and instantly edit satellite imagery — turning real buildings, roads, and borders into whatever the prompt asked for. Within a day, the feature was gone. According to The Verge, a researcher used it to generate images of refugees near the Mexican border and a bomb crater beside a hospital in Gaza. Google rolled the feature back, admitting people “uniquely trust Google Earth for a reliable view of the world.”
Before you file this under “tech company drama,” look at what actually happened: the world’s most trusted map platform let people rewrite real geography with a sentence. And if that’s possible on Google Earth, it’s possible anywhere — including in the images your customers, suppliers, and competitors see of your business every day.
TL;DR
Google launched an AI feature that let anyone edit real-world satellite images with text prompts. Within 24 hours, people had used it to fabricate scenes of conflict, and Google pulled the entire tool. The takeaway for your SME: AI-generated images are now effectively indistinguishable from reality, and your business’s trustworthiness has become your most valuable filter in a world where fakes are effortless.
What This Means
A “deepfake” used to mean a video of a person saying something they never said. This incident shows the definition has expanded to the physical world itself. The tool, powered by an image model called Nano Banana 2, could take a real satellite image and produce an altered version so believable you’d act on it — check on your property, file a complaint, change a route, verify a delivery.
Google’s initial response had two defences: generated images carried a digital watermark, and the system claimed to block “image creation on harmful topics.” Neither held up. Researcher Henk van Ess showed that AI detection tools could be fooled, and reported that “nothing was refused, nothing was softened.” The watermark existed, but it didn’t stop anyone from screenshotting and sharing.
Watermarks and policy rules only matter if people check them. The moment a fake image looks real enough to share, the screenshot becomes the truth — watermark or not.
The 24-hour timeline
According to The Verge’s reporting, here’s how quickly it unravelled:
| When | What happened |
|---|---|
| Day 1 | Google launches AI-based image editing in Google Earth, powered by Nano Banana 2 |
| Day 1 | Researcher Henk van Ess generates fake images of refugees near the Mexican border and a bomb crater by a Gaza hospital |
| Day 1 | Google says watermarks and harmful-topic filters are enough; van Ess shows AI detection tools can be fooled |
| Day 2 | Google rolls back the feature entirely, saying people trust Google Earth for a reliable view of the world |
How This Applies to Malaysian SMEs
Consider how many times a day your business depends on images being real. If you run a restaurant or F&B business, your menu photos on Grab Food or Shopee are the reason customers order. If you’re in property, your listing photos on iProperty and Mudah.my are the reason buyers show up for a viewing. If you sell products online, your product images are the entire sales pitch. Now picture a competitor — or a disgruntled customer — generating a fake image of your product, your shopfront, or your food and sharing it publicly. You’d spend days undoing something you never did.
The risk is even closer on WhatsApp. Malaysian business groups and neighbourhood chats treat forwarded images as evidence. Fake photos of floods, accidents, or viral “health hazards” spread faster than official announcements. If you’re in construction or renovation, a client could submit an AI-faked image of damage to back a dispute. If you run a delivery or logistics operation, a supplier could present an edited image as proof of completed work. Google Earth was the last source you’d expect to generate fiction — and it did, freely, for 24 hours.
Here’s the upside. As AI-generated images become widespread, verifiable authenticity becomes a competitive edge. Malaysian consumers are already sceptical of over-edited food photos and retouched property listings. A business that clearly labels its AI-assisted visuals and shows the real, unedited originals wins trust in a market flooded with polish. Transparency isn’t a nice-to-have anymore; it’s the simplest way to stand out.
You can also prepare your internal processes. Keep camera originals with timestamps. Record video walkthroughs of your premises or stock. If an image ever becomes part of a dispute — with a customer, supplier, or even a government agency — the original file is your evidence. Everything generated or edited should be stored separately and labelled.
Practical Takeaways
- Verify before you act. When someone sends you an image as proof — of damage, delivery, or completed work — ask for the original file and look for editing clues before you respond.
- Label your AI-assisted content. A simple “enhanced with AI” note on marketing images builds a reputation for honesty before anyone can catch you out.
- Keep raw files. Store original camera photos and videos with timestamps. In a dispute, the unedited original is worth more than any argument.
- Set a simple internal rule. If your team uses AI tools — for ads, product photos, or social posts — require a label and a saved original for every output.
- Stay informed. AI tools are being embedded into everyday software faster than policies can keep up. Make checking for new AI features in your current apps a monthly habit.
The Bigger Picture
This one-day experiment is a preview of a bigger change. AI image generation is moving into mainstream tools — maps, photo editors, design apps, even your phone’s camera. The question isn’t whether your industry will see AI-faked images. It’s when, and how ready you’ll be.
For Malaysian SMEs, the practical response isn’t to avoid AI — it’s to build verification into your workflow the same way you built invoicing and inventory systems. Real photos, original files, and clear labelling take minutes to set up, and they protect you from the kind of trust collapse Google experienced at global scale in 24 hours.
If you want to stand out in the next decade, make “real” your brand. The tools to fake things are getting better every month. The tools to prove things are real are still simple: originals, timestamps, and honesty.
Ready to Streamline Your Operations?
Your business should run itself. AutoRunBiz deploys AI agents to automate your daily operations — WhatsApp orders, invoicing, customer follow-ups, and accounting. Book a free 15-min ops audit to see where automation fits your business →