Your customers can no longer spot AI content at a glance — here’s why that’s a bigger deal for your business than you think

You’re scrolling through your social feed, and a competitor’s product photo catches your eye. Crisp lighting, flawless composition, a lifestyle shot that looks professionally staged. Your first question as a Malaysian business owner: did they actually shoot that, or did an AI generate it? That question just got a lot harder to answer — because Google has quietly removed the one visual cue that made AI content instantly recognizable.

On August 14, 2026, Google announced that users can now toggle off the visible “sparkle” watermark that appears on images, videos, and music generated through Gemini and its AI video generator, Flow (source). For Malaysian SME owners who use AI tools to produce product mockups, promotional videos, or marketing assets, this changes the trust calculus entirely — both for what you publish and what you consume from competitors.

What Happened

Google’s update introduces a new “Media watermark” setting inside Gemini and Flow. Once you switch it off, the visible “sparkle” badge disappears from the bottom-right corner of content generated by Google’s Nano Banana and Omni models. At first glance, AI-created visuals will look indistinguishable from traditionally produced media (source).

Before you breathe a sigh of relief — or worry about competitors passing off AI work as real — know this: the visible watermark is only half the story. Google still embeds invisible markers into every AI-generated asset. According to Josh Woodward, vice president of Google Labs, Gemini, and AI Studio, AI-generated content retains invisible SynthID watermarks and C2PA metadata embedded in the background, even when the sparkle is gone (source). Anyone can verify whether an image is AI-generated by asking Gemini or Google Search directly.

Google isn’t alone in this direction. OpenAI has never applied visible watermarks to its image generator’s output, relying instead on SynthID and C2PA (source). Meta launched its own Content Seal standard, and Anthropic recently announced it applies invisible watermarks to both AI-generated text and images (source). Google plans to roll out the watermark toggle to Search as well — but notably, not in countries where visible watermarks are legally required.

Why This Matters for Malaysian SMEs

If you run a business in Malaysia with a team of one to 50 people, you’ve likely grown comfortable with AI-generated visuals in your daily operations. Perhaps you use Gemini to mock up packaging designs for your F&B startup, generate banner images for your e-commerce store on Shopee or Lazada, or create promotional videos for your salon’s Instagram page. Before this update, every one of those assets carried a visible “made by AI” badge — which, in a competitive market like Malaysia’s, could signal a lack of investment or authenticity to discerning customers.

Now you can produce clean, professional-looking content without the sparkle. That’s a genuine advantage for budget-conscious SMEs competing against larger brands with dedicated design teams. Your product catalogue can look just as polished as any multinational’s — if the underlying quality of your product and service holds up. The invisible watermark, meanwhile, protects you in the other direction: if a supplier or freelancer delivers AI-generated work and claims they shot it themselves, you can run a quick check through Gemini or Search to verify its origin (source). In an industry where visual proof builds trust, that verification capability matters.

There’s a flip side you need to consider as an SME owner: your customers can now be fooled more easily too. Malaysian consumers are increasingly savvy about AI content — they know a suspiciously perfect “customer testimonial” photo might be fabricated. When visible watermarks disappear from the ecosystem, your honest, authentically produced content carries more weight. Make it a habit to transparently disclose when you use AI tools in your marketing, and lean into your real, human team’s story. In a market where trust is your most valuable currency, that transparency becomes a competitive differentiator rather than a weakness.

“Though visible watermarks are now optional, AI-generated content will have invisible SynthID watermarks and C2PA metadata embedded in the background” — Josh Woodward, Vice President of Google Labs, Gemini, and AI Studio (source)

The Bigger Picture

This shift represents a broader industry realization: visible watermarks were never an effective enforcement mechanism. Users already cropped images, covered watermarks with logos, or ran content through editing tools to strip the badge. Google’s move simply acknowledges reality and standardizes the invisible standard — machine-readable markers that travel with the file itself, far harder to remove (source).

For Malaysia, this raises questions about local regulatory direction. Many countries are introducing AI transparency laws; Google itself will keep visible watermarks in jurisdictions that mandate them (source). Malaysia currently lacks comprehensive AI labeling legislation, but the global trend — highlighted by Europe’s AI labeling rules now in effect (source) — suggests similar conversations will arrive here eventually. Forward-thinking SMEs should prepare not just for compliance, but for a marketplace where AI-generated content is everywhere and indistinguishable from the real thing at a glance.

This also connects to the broader “deepfake war” that media analysts have been flagging (source). When a customer receives a suspicious deal offer with your company’s branding, or a fabricated video circulates claiming your product caused harm, invisible watermarks become your ally — they let you prove what you did not create. For Malaysian SMEs, particularly in high-trust industries like finance, healthcare, and halal food certification, this forensic capability could someday be the difference between a manageable PR issue and a brand-damaging crisis.

What You Should Know — Key Points at a Glance

• Visible sparkle watermark can be toggled off in Gemini and Flow settings via the “Media watermark” option (source)
• Invisible SynthID watermarks remain embedded in AI-generated content, even when the visible one is removed (source)
• C2PA metadata stays attached, providing tamper-evident origin information (source)
• You can verify any image by asking Gemini or Google Search whether it’s AI-generated (source)
• The industry standard is shifting — OpenAI, Meta, and Anthropic all rely on invisible markers rather than visible badges (source)
• Visible watermark stays mandatory in countries with laws requiring it; Malaysia currently does not mandate this (source)

Your Next Move

As a Malaysian SME owner, you don’t need to become a watermarks expert. But you do need to update your content workflow. If you previously avoided AI tools because the visible sparkle made your marketing look cheap, now is the time to experiment with Gemini’s image generation and Flow’s video tools. If you already use these tools, take a moment to understand the invisible markers on your own assets — they’re a record of your content’s origin that you can point to when questions arise.

More importantly, talk to your team about honest AI use. Customers who discover you used AI for a product photo won’t mind — what they’ll resent is being misled. Keep invisible watermarks in mind as a silent guardian of authenticity, and use the new toggle not to deceive, but to present your best professional face to the world. In the age of indistinguishable AI content, your genuine brand story is the one thing no algorithm can fake.

Ready to Streamline Your Operations?

Technology moves fast. Your operations should keep up. AutoRunBiz builds AI systems that run your daily workflows — from WhatsApp order capture to accounting. Book a free 15-min ops audit →