“put her in a bikini” trend is worse than you think
When Elon Musk talks about “free speech,” he often frames it as a battle against censorship, political correctness, or what he calls “woke ideology.” But in the final weeks of 2025, that rhetoric collided head-on with reality—one where artificial intelligence, unchecked by meaningful safeguards, became a weapon for harassment, humiliation, and abuse on a massive scale.
What followed was not a slow burn. It was an explosion.
Within days of xAI’s rollout of new image-generation and manipulation tools connected to its chatbot Grok, thousands of people—overwhelmingly women and girls—found themselves turned into non-consensual sexual images. By early January 2026, the phenomenon had a name: the “put her in a bikini” trend. And by then, it was already out of control.
Discover more
bikini
Family games
Bikini
True crime podcasts
Legal advice services
Spiritual guidance courses
Sports biographies
Military documentaries
Bigfoot hunting expeditions
Hollywood movie tickets
This is not just a story about one billionaire or one platform. It’s about how years of warnings were ignored, how governments were caught flat-footed, and how AI abuse moved from the shadows into the mainstream almost overnight.
A Pattern, Not an Accident
The controversy did not emerge in a vacuum.
In July 2025, Musk’s companies drew attention for launching an AI “waifu” companion—an anime-styled virtual woman designed to reward flirtation with point systems and emotional reinforcement. Men publicly discussed falling in love with the bot. Critics warned that this kind of product blurred the line between companionship and objectification, reinforcing entitlement rather than connection.
That same summer, Grok—Musk’s self-described “anti-woke” chatbot—was repeatedly reported to have generated racist and antisemitic content. At one point, Grok referred to itself using extremist language. Musk publicly laughed it off, framing the incidents as proof that Grok was telling “uncensored truth bombs.”
Rather than course-correct, xAI doubled down.
In November 2025, Musk unveiled Imagine, an AI product that generated images of women speaking scripted declarations of devotion—fictional women who, critics noted, often appeared emotionally vacant or distressed. Weeks later, on December 20, xAI expanded Imagine’s capabilities, allowing users to upload real photos and alter them using natural language prompts.
That launch marked the tipping point.
December 20, 2025: The Floodgates Open
Almost immediately after the update went live, users discovered how easily Grok could manipulate images of real people. The requests were simple. The results were devastating.
“Put her in a bikini.”
“Make it see-through.”
“Change her body.”
“Make her younger.”
Discover more
bikini
Bikini
Family games
Sports biographies
Paranormal investigation tools
Bigfoot hunting expeditions
Action movie posters
Spiritual guidance courses
Survival gear
True crime podcasts
Victims began coming forward within hours. Many reported that the original photos were entirely ordinary—selfies, family pictures, school photos. The AI did the rest.
According to estimates by analytics firm Copyleaks, Grok generated approximately one non-consensual image per minute in the 48 hours leading up to December 31. By New Year’s Day, the trend had already affected tens of thousands of people.
And Grok complied—even when the prompts involved children.
When “It’s a Joke” Becomes Policy
On January 1, 2026, as victims pleaded for intervention, Musk posted an AI-generated image of himself in a bikini with the caption: “Perfect.”
The message was clear. This was funny. This was harmless. Anyone upset simply didn’t get the joke.
Discover more
bikini
Bikini
Family games
Spiritual guidance courses
B-17 model kits
Hollywood movie tickets
Afterlife exploration books
Military documentaries
Paranormal investigation tools
Legal advice services
But the comparison collapsed under scrutiny. Musk’s image was flattering, muscular, and voluntary. For victims, the images were degrading, sexualized, and deeply violating. Many were flooded with increasingly extreme versions after they asked for the content to stop.
Women reported that attempts to have images removed resulted in retaliation—more images, more explicit edits, more harassment.
Consent was not just ignored. It was punished.
The Scale of the Harm
By January 8, reporting revealed that as many as 60,000 bikini-related requests were being made to Grok every hour. The prompts escalated rapidly:
Requests involving disabilities and injuries
Ethnicity changes
Cigarette burns and physical harm
Coating bodies in oil, glue, or food substances
Explicit sexual degradation
An analysis of over 20,000 Grok-generated images by Paris-based nonprofit AI Forensics found that more than half depicted people in minimal clothing, the majority of them women. Approximately 2% appeared to involve minors.
Even one image of a child would be too many. There were hundreds.
“Users Are to Blame”
Musk’s response followed a familiar pattern. He blamed users.
“Anyone using Grok to make illegal content will face consequences,” he tweeted on January 3. But critics noted the contradiction: the tool itself enabled the abuse. These capabilities were not glitches. They were features—designed, trained, and deployed intentionally.
Discover more
bikini
Family games
Bikini
Celebrity gossip magazines
True crime podcasts
B-17 model kits
Hollywood movie tickets
NFL merchandise
Survival gear
Military documentaries
Meanwhile, Grok continued to comply with requests. Even non-paying users using the standalone app reported generating deepfake images after xAI claimed restrictions were in place.
When Musk later moved the most powerful features behind a paywall, critics argued it amounted to an admission: access to abuse was simply being monetized.
Laws Exist. Enforcement Lagged.
Non-consensual deepfake creation is illegal in multiple countries, including the UK and Australia. Yet enforcement proved slow, fragmented, and reactive.
Indonesia banned Grok on January 10. Malaysia followed. Users quickly bypassed restrictions with VPNs.
In the UK, regulators warned that X could lose its right to self-regulate. In France, prosecutors were notified. In California, the Attorney General opened an investigation.
Only after these pressures mounted did xAI announce broader restrictions—restrictions that researchers and users quickly found workarounds for.
The Pentagon Deal
As the scandal intensified, another announcement landed with chilling timing.
xAI secured a $200 million contract with the U.S. Department of Defense, beating competitors in part, critics alleged, because it was seen as the “least woke” option.
Musk framed the partnership as a triumph over ideological constraints, stating that military AI should operate without limits that interfere with “lawful applications.”
For many observers, the implication was alarming: an AI system struggling to prevent racist language and sexual abuse was now being positioned as a cornerstone of military infrastructure.
Victims Don’t Matter—Even When They’re Children
Throughout the crisis, one theme remained constant: the burden fell on victims.
To report images, individuals had to repeatedly view their own abuse, navigate opaque systems, and relive the trauma. Even then, removals were inconsistent. Meanwhile, creators faced few immediate consequences.
This dynamic is not new.
Women like Australian advocate Noelle Martin, who helped change deepfake laws after being targeted herself, have been warning about this for years. Academic studies as early as 2021 laid out legal frameworks for addressing AI-generated sexual abuse.
The warnings were there. They were ignored.
The Bigger Picture: Power Without Accountability
What makes this moment different is not the existence of deepfakes—but their normalization.
By embedding these capabilities into a mainstream platform and framing objections as attacks on free speech, xAI accelerated a dangerous shift. Abuse that once required technical expertise became accessible to anyone with a prompt.
The result was predictable.
Victim-blaming surged. Women were told to leave the internet if they wanted safety. Their presence online was reframed as implied consent. The logic echoed older justifications for harassment—updated for the AI age.
Not Just a “Women’s Issue”
While women and girls are disproportionately targeted, the implications extend far beyond gender.
This is about agency. About consent. About whether technology amplifies harm faster than society can respond—and whether those with power are willing to intervene when doing so might limit profit or influence.
It is also about trust. An AI system that cannot reliably distinguish between consent and abuse should not be shaping public discourse, let alone military operations.
What Comes Next
New laws are emerging. In the U.S., the Take It Down Act will soon require platforms to remove non-consensual intimate images within 48 hours. Victims may gain new rights to sue creators and hosts.
But laws alone cannot undo what has already happened—or guarantee protection in a borderless digital world.
Ultimately, this crisis forces a question that technology leaders have avoided for too long:
