AI is no longer just writing bad poetry and faking celebrity voices. It’s being used to recreate real crimes — and the victims are getting hurt twice. This is the moment the tech industry’s “move fast” mentality collides head-on with human suffering, and someone needs to say it out loud.
South Korea is staring down a crisis that most Western tech commentators haven’t fully registered yet. According to a report from the Chosun Ilbo, AI-generated videos depicting criminal acts are circulating online and triggering what experts are calling “secondary harm” — a clinical term for something deeply ugly. Victims of crimes are watching fictionalized, AI-reconstructed versions of their worst moments spread across social media. Over and over. Without consent. Without end.
This isn’t hypothetical anymore. It’s happening now.
The Machine Has No Conscience
Here’s what makes this different from every other AI content panic we’ve had before. This isn’t deepfake porn, though that’s catastrophic enough. This isn’t a politician’s voice cloned for a prank. This is the re-dramatization of violence — murders, assaults, kidnappings — rendered in hyper-realistic AI video and served up to audiences hungry for true crime content.
The platforms hosting this content aren’t evil masterminds. They’re just indifferent. Indifference at scale is its own kind of cruelty.
Families of murder victims are reportedly encountering AI videos that simulate how their loved ones died. Think about that for three full seconds. These aren’t documentaries. They’re not journalism. They’re content — produced to generate clicks and watch time — dressed up in the grammar of information.
True Crime Has Always Had an Ethics Problem
Let’s be honest. True crime content has always walked a morally uncomfortable line. Podcasts, Netflix docs, Reddit threads — we’ve been consuming real human tragedy as entertainment for decades. The genre exists in a permanent ethical gray zone where public interest and voyeurism are almost impossible to separate.
AI just removed the last speed bump. Before, you needed resources, access, and at least a veneer of journalistic purpose to produce something that looked like a documentary. Now? You need a prompt and a free account. The barrier to recreating someone’s worst day has dropped to nearly zero.
That’s not progress. That’s a trap with a slick interface.
Secondary Harm Is the Real Story
Criminologists and trauma specialists have documented secondary victimization for years — the way survivors get re-traumatized by media coverage, court proceedings, or public curiosity. AI-generated criminal content industrializes that process. It’s secondary harm on demand, served algorithmically to anyone who clicks.
South Korea’s response has been faster than most countries would manage. Regulators there are actively pushing for clearer legal definitions around AI-generated content that depicts real crimes. The conversation includes whether this falls under defamation law, privacy law, or something entirely new that existing frameworks can’t handle.
Spoiler: existing frameworks can’t handle it. They were written before a teenager with a laptop could generate photorealistic crime scene reconstructions in twenty minutes.
The Platforms Will Not Save You
YouTube, TikTok, and Meta all have content policies that technically prohibit content that glorifies violence. In practice, enforcement is inconsistent at best and performative at worst. AI-generated content makes moderation exponentially harder because the tells are disappearing fast. A video that would have looked fake eighteen months ago now passes casual inspection.
We wrote recently about how new carbon-free hydrogen fuel production methods are slashing temperature requirements by 900°F — a reminder that technology, when pointed in the right direction, can solve enormous problems. The same engineering brain trust producing miracle energy solutions is also producing tools with zero guardrails that are actively hurting people. The contrast is not subtle.
Platform self-regulation has been the industry’s answer to everything for fifteen years. That answer has aged badly. It was never enough for hate speech. It wasn’t enough for misinformation. It is absolutely not enough for AI-generated criminal content targeting real victims.
The Hot Take
AI-generated content depicting real crimes should be treated as criminal harassment — full stop, no exceptions for “artistic interpretation.” Yes, that means aggressive legislation. Yes, that will be imperfect and occasionally overreach. That’s still better than the alternative, which is a world where any grieving family can stumble across a pixel-perfect simulation of their child’s death on a Tuesday afternoon. Free speech is not a suicide pact with the vulnerable.
What Actually Needs to Happen
Legislators need to stop being impressed by tech founders in Senate hearings and start writing laws with teeth. Victims’ rights organizations need a seat at the table in AI policy discussions — not as an afterthought, but as primary stakeholders. And the rest of us can start by making a choice about what kind of habits we build around the content we consume — because what we click on funds what gets made.
The technology is not going back in the bottle. The question is whether society builds walls fast enough to protect the people most likely to be hurt by it. Right now, the answer is no — and the victims are paying that price in real time, every single day, while the rest of us debate the finer points of AI ethics at conferences with open bars.