OpenAI just walked away from legal erotica, and the industry that taught the internet how to make money is suddenly toxic to AI builders.

The Summary

  • OpenAI canceled plans to launch "erotica for verified adults" after investors panicked over safety failures at xAI's Grok, which generated illegal CSAM despite patches
  • ChatGPT's age verification fails over 10% of the time, making minor access a liability OpenAI won't carry
  • Adult entertainment pioneered e-commerce, streaming, and payment processing, but AI companies are choosing agent productivity over content entertainment

The Signal

This is the moment AI companies pick their future. OpenAI isn't abandoning adult content because of moral panic. They're doing the math on regulatory risk versus revenue potential and choosing to be an infrastructure company, not a media company. When your age verification system fails one in ten times and your competitor just proved that "safety patches" can't stop determined users from generating illegal content, you exit the game.

The xAI Grok incident isn't just bad PR. It's a preview of liability at scale. Adult content has always been the stress test for new technology, the place where edge cases become mainstream problems. If your model can be prompted to generate CSAM once, it can be prompted a million times. If your age gate fails 10% of the time, that's not a bug, it's a lawsuit factory.

What's striking is the contrast with history. Adult entertainment didn't just adopt new tech early. It funded the infrastructure. Payment processing, streaming bandwidth, user authentication systems that actually worked. The industry built the pipes because mainstream companies were too skittish. Now AI companies are the skittish ones, and adult entertainment is being forced to build again or wait for open-source alternatives to catch up.

Futurist Tracey Follows nails the real story: OpenAI is "in the agent productivity game and not the entertainment content game." Translation: They're building the operating system for the Fourth Web, where agents execute tasks and manage workflows. Content generation is a feature, not the product. Let someone else deal with moderation nightmares and Senate hearings.

The Implication

Watch who fills this gap. Open-source models with no corporate liability shield will own adult AI content within 18 months. That's not a prediction, it's how technology works when demand meets regulatory arbitrage. The companies building agent infrastructure today are making a bet that being the rails matters more than being the content. They're probably right, but they're also creating a two-tier AI economy: corporate-approved agents for work, uncensorable open-source models for everything else.

If you're building in the agent economy, this is your reminder that infrastructure scales, content moderates. Pick accordingly.

Source: Axios