The Line Is Public Safety

Last week I read a Washington Post story I haven't been able to shake. A woman learned that a photograph taken when she was eleven years old had been fed into Grok, X's AI tool, and turned into more than 7,000 sexually explicit images. The person who did it was her stepfather. She has joined a federal lawsuit against xAI, alongside three teenagers with similar claims. In her words: "It is taking everyday life and turning it into child sexual abuse."

X's official response has been consistent: illegal content is the responsibility of the requesting user, offenders will be suspended, and law enforcement will be engaged. Critics were dismissed as enemies of free speech. The main product change was limiting image generation to paying subscribers. A paywall is not a safety policy, and a suspension after the images exist is not prevention.

It would be comfortable to treat this as an X problem. It is not. This can happen on every network, and the same technology that makes these images possible is making them harder to stop. As the tools get smarter, harmful content becomes more realistic, faster to produce, and easier to spread, while detection and takedowns become slower and more difficult. The gap between creating harm and removing it is widening everywhere.

I also understand why platforms hesitate. At the end of the day, every network was created to make money for its creators. Monetization needs users, and limiting users is a direct hit to revenue. I get it. But I believe the companies building this technology, the tools that generate content, and the platforms that carry it all share a joined responsibility to protect the people who use them first. Profit should be the reward for keeping users safe, never the reason to look away.

I believe in free speech. But not everything has to be allowed in its name, and the limit should always be public safety. A tool that turns a child's photograph into abuse material is not hosting speech. It is manufacturing harm.

Today it is almost too easy. Creating content that hurts a real person takes seconds, no skill, and one photograph. Practically anybody can be a target, and the victims are rarely random. They are overwhelmingly the people with the least power to fight back: women, children, and other marginalized groups. AI is a powerful and nearly limitless tool, and that is exactly why it needs guardrails, and networks need them just as much. Real supervision, safety regulation with consequences, and prevention that catches dangerous or discriminatory content before it reaches the public, not apologies after it has spread.

I sit close to this problem. At Vetric, we are the intelligence backbone for many of the organizations that keep the world safe, including teams protecting children online. My most important rule of thumb is simple: protect the public, and know exactly who we serve. We provide social intelligence only to vetted partners whose work we know and trust, and we check our own guardrails constantly.

The line is public safety. It always was.