A new viral trend sweeping X, known as “kakushie” (from the Japanese word for “hidden”), is exploiting the platform’s image compression to create optical illusions that have captivated millions of users, but the same mechanism has also raised alarms about potential misuse to conceal disturbing material behind innocuous-looking posts.
Как работает этот трюк
The trend, which gained momentum in late July 2026, involves creators uploading specially prepared transparent images that appear faint, washed-out, or nearly blank in a user’s timeline.
When someone long-presses the image on mobile, triggering the platform’s high-resolution loading option, the full-quality original renders against the app’s dark image viewer, revealing vivid details that weren’t visible in the compressed preview.
The effect relies on a technical quirk: X compresses timeline thumbnails in a way that strips out certain transparency data, but the full-resolution version loads differently against the viewer’s dark background, making the same file look dramatically different depending on how it’s displayed.
The trend traces back to Japanese art communities on X, where illustrators originally used the effect to show a rough sketch in the feed and the finished artwork on hold. Since then, dedicated tools have sprung up online to help users generate their own versions of the effect.
The trick has proven a potent engagement driver, since it effectively forces users to interact with a post to see what it actually contains, a behavior X’s recommendation algorithm tends to reward with wider visibility. Posts using the effect have racked up millions of views within hours, according to creators tracking the trend’s spread.
A Platform Already Under Scrutiny
The appeal of the trend is also its risk: the same gap between what a compressed preview shows and what the full-resolution image reveals could, in principle, be used to hide content that wouldn’t pass moderation if shown plainly.
Some users have flagged concern that the technique could be used to mask explicit material behind ordinary-looking thumbnails, including the kind of sexualized imagery platforms are already struggling to catch under normal moderation. X has not addressed these specific concerns publicly.
The worry lands at an already difficult moment for the platform’s content moderation record. A June 2026 investigation by El País found it was “still extremely easy to find child sexual abuse material on X,” with certain hashtags flooded by accounts spreading such content.
The UK’s Ofcom has a separate, ongoing investigation into whether X met its obligations under the Online Safety Act after reports that its Grok chatbot had been used to generate undressed images of real people and sexualized images of children, an inquiry Ofcom called a “matter of the highest priority.” In January, US senators separately pressed X for details on its protections against sexualized AI-generated imagery on the platform.
X introduced manipulated media warning labels earlier this year, but those were built with political misinformation in mind rather than the kind of visual obfuscation this new trend relies on, and it’s unclear whether existing moderation tools are equipped to catch content hidden this way before it spreads.
Быстрые ссылки:
