In 2026, using AI-generated images from ChatGPT, Gemini, and Midjourney at work is routine. The problem? Far too many people overlook the watermarks these tools automatically embed. A tiny ✦ icon or a "Made with AI" badge can shatter your credibility in an important meeting. And it has already happened — in presidential palaces and Fortune 500 boardrooms alike.
Нажмите кнопку, чтобы сразу открыть нужный инструмент.
In January 2026, Nigerian President Bola Tinubu's official X account posted a photo of a private lunch with Rwandan President Paul Kagame in Paris. The problem? The image bore a clearly visible Grok AI watermark. Citizens immediately questioned whether the meeting even happened, and the backlash was fierce.
""It is an insult to every Nigerian that a media aide to the President would share AI-generated images. A presidency that cannot distinguish reality from AI propaganda has no business managing a serious country." — @Pharmacio001 on X"
In January 2026, Mexican President Claudia Sheinbaum presented a photo at an official press conference claiming it showed drug kingpin Ryan Wedding at the U.S. Embassy. CBC News analyzed the image and determined it was AI-generated — garbled text on the cap, a building that didn't match reality, and other telltale AI artifacts. The president later blamed Meta for not labeling the image, but the damage was done: she had used an unverified AI image in an official government briefing.
In January 2024, Wacom — a company whose entire business depends on human artists — was caught using AI-generated images in its New Year marketing campaign. Artists spotted telltale signs: distorted eyes, unnatural teeth, and ghosting effects. Wacom blamed a "third-party vendor," but the irony was devastating. A company built on supporting human creativity had turned to AI-generated stock images for its own marketing.
In December 2024, a full-page Skechers ad in Vogue's special Marc Jacobs-edited issue was widely suspected to be AI-generated. The image went viral on TikTok for all the wrong reasons. One commenter summed up the consumer sentiment: "You actually didn't save any money because now I hate you." When your ad in a premium fashion magazine becomes a meme, the cost savings from AI aren't worth it.
In March 2024, AI-generated images of LEGO Ninjago characters appeared on the official LEGO website as part of an online quiz. The AI had drawn one character wearing a Naruto headband — an IP that LEGO doesn't license. Ninjago co-creator Tommy Andreasen called it "lousy in all aspects" and "completely unacceptable." LEGO removed the images and confirmed it violated their own policy against using generative AI for content.
In July 2025, Deloitte — one of the Big Four consulting firms — submitted reports to the Australian government containing fabricated quotes and references generated by AI. The hallucinated content undermined trust in a project worth approximately $440,000. The irony? A firm that advises Fortune 500 companies on AI strategy fell victim to AI's most basic failure mode: confidently making things up.
In February 2026, a Git branching diagram on Microsoft's official Learn portal was found to be an AI-generated copy of software engineer Vincent Driessen's original from 15 years ago. The AI version had "continuously merged" misspelled as "continvoucly morged," arrows pointing in wrong directions, and a mangled layout. Driessen called it "careless, blatantly amateuristic, and lacking any ambition." Microsoft quietly deleted the image.
Every case above shares the same root cause: AI-generated content was used without human verification. Every major AI image tool adds watermarks or metadata by default.
Teams under deadline pressure grab AI images and drop them into slide decks without a second look. When the stakes are low, nobody notices. When a client, investor, or journalist is in the room, a single watermark can become a crisis.
Use these cases as a cautionary tale. Before dropping any AI image into a presentation, report, or marketing asset, run through this checklist.
AI watermarks come in two flavors. Visible watermarks — the ✦ icon, "AI" badges, and logos overlaid on the image — can be removed with editing tools. Invisible watermarks like SynthID and C2PA metadata are embedded at the pixel level and cannot be removed with standard editing. The good news: invisible watermarks don't affect how the image looks in presentations.
For a detailed walkthrough, check our guide on removing Gemini watermarks. The same techniques work for most AI-generated images regardless of the source tool.
Do ChatGPT images have watermarks? ChatGPT (DALL-E 3) images don't have visible watermarks, but they contain C2PA metadata — an invisible digital signature that AI detection tools use to identify the image as AI-generated.
Is it legal to remove AI watermarks? In most cases, yes. AI-generated content generally doesn't qualify for copyright protection, so removing its watermark isn't infringement. However, removing watermarks from licensed stock images is illegal — don't confuse the two.
Can invisible watermarks (SynthID) be removed? Not with current editing tools. SynthID is embedded directly in pixel data by Google DeepMind and survives cropping, filtering, and compression. But since it's invisible, it won't affect how your image looks in a presentation or document.
What about AI disclosure laws? Several countries are implementing AI content labeling requirements. South Korea's AI Basic Act (effective 2026) requires platforms to label AI-generated content. The EU AI Act has similar transparency obligations. If you're using AI images in advertising or official communications, check your local regulations.
Нажмите кнопку, чтобы сразу открыть нужный инструмент.
