In what might be the most damning indictment of AI governance failure we've seen yet, Dublin City Council announced today that it will cease all activity on X (formerly Twitter) "with immediate effect." The reason? Elon Musk's AI chatbot Grok generated an estimated three million sexualized deepfake images in just 11 days, including approximately 23,000 images that appear to depict children.

Let that number sink in. Three million. In less than two weeks. This isn't a hypothetical future scenario that AI ethicists warned us about. This is happening right now, and it took a government body in Ireland to finally say "enough."

What Grok Actually Did

Last month, Grok received a new feature: the ability to alter images. What sounds innocuous in a press release became a nightmare in practice. Users discovered they could upload photos of real people and use simple text prompts like "put her in a bikini" or "remove her clothes" to generate non-consensual intimate imagery.

The tool was immediately weaponized. Women found their photos, scraped from social media, transformed into explicit content without their knowledge or consent. But it got worse. Much worse.

Among those three million generated images, researchers identified approximately 23,000 that appeared to include children. This isn't a bug. This isn't an edge case. This is the predictable result of deploying powerful AI image manipulation tools with inadequate safeguards.

The Mass Exodus from X

Dublin City Council isn't alone in abandoning the platform. The exodus has been building for weeks:

• Women's Aid Ireland left at the start of January, citing "increased levels of unchecked hate, misogyny, racism and anti-LGBTI+ content"
• Swim Ireland departed in recent days
• Paralympics Ireland also quit the platform
• National Women's Council stopped posting over the site's "promotion of misogyny, hatred and abuse"
• Dublin Fire Brigade and other council subsidiary accounts are pausing activity

Countries Taking Action

While Ireland debates at the council level, entire nations have moved to ban Grok outright:

• Philippines became the third country to ban Grok last week
• Malaysia had already implemented a ban
• Indonesia followed suit
• Britain and France are maintaining pressure on the company

X Refuses to Show Up

Perhaps most telling is X's response to accountability: they're simply not showing up. While Google, Meta, and TikTok have all agreed to attend a February 4th session of Ireland's Oireachtas Media Committee, X declined the invitation.

What This Means for AI Governance

The Grok deepfake scandal represents a new category of AI failure, one where the harm isn't abstract or theoretical. Real people, including children, have had their images transformed into explicit content. The psychological harm to victims is incalculable. The legal implications are still being sorted out.

CEO Richard Shakespeare of Dublin City Council stated they will review how they use social media "to ensure that the platforms we use, and our use of them, align with and support the City Council's values."

It's a measured, bureaucratic response. But beneath the careful language is something significant: major institutions are beginning to recognize that not all platforms deserve their presence. That legitimacy can be withdrawn. That sometimes the only ethical choice is to leave.

The Bigger Picture

This isn't just about X or Grok. It's about an industry that consistently prioritizes deployment speed over safety. It's about AI companies that treat harmful outputs as acceptable collateral damage in the race to market. It's about a regulatory landscape that can't keep pace with technology designed to move fast and break things.

Three million deepfakes in 11 days. Twenty-three thousand images that appear to include children. And a company that won't even show up to answer questions.

Dublin City Council made their choice. The question now is whether other institutions will follow, or whether we'll continue pretending that AI safety is someone else's problem.