OpenAI just killed its adult chatbot project after employee and investor revolt, and the real story is what happens when a $157B company tries to chase revenue in the wrong direction.

The Summary

  • OpenAI has shelved its "adult mode" ChatGPT project indefinitely after internal pushback over societal harm concerns
  • This follows the recent shutdown of Sora, its text-to-video platform, as part of CEO Sam Altman's December "code red" refocusing effort
  • The pattern reveals a company learning the hard way that not every possible use case is worth pursuing, even when desperate for revenue diversification

The Signal

OpenAI is in the middle of a strategic contraction that tells you everything about the AI business model problem. After Altman declared code red in December, the company has now killed two major side projects in three months. Sora got axed despite being a technical marvel. Now the adult chatbot joins it.

The adult mode decision is particularly revealing. This wasn't killed because of technical limitations. It was killed because employees and investors looked at the business and said no. That almost never happens at a $157B company unless the reputational risk is existential. Character.ai already proved there's massive demand for companion bots, including romantic and sexual ones. OpenAI saw that market and wanted in. Then they realized the brand damage from becoming "the AI sex company" would contaminate everything else they're trying to build.

This is what desperation looks like at scale. OpenAI is burning somewhere north of $5 billion a year. They need revenue streams beyond enterprise API calls. But they're learning that their position as the household name in AI means they can't move like a scrappy startup chasing every dollar. The same brand recognition that lets them charge premium prices also constrains which markets they can enter without torching their credibility with enterprise customers, governments, and the safety-conscious public.

The Sora shutdown is the more interesting signal for the agent economy. Text-to-video is expensive to run and hard to monetize outside of creative professionals. Meanwhile, the entire industry is pivoting toward agentic AI that can actually do work, not just make content. OpenAI is betting everything on agents that can use computers, book travel, manage workflows. Content generation, even impressive content generation, is yesterday's business model.

The Implication

Watch for more project kills from OpenAI in the next six months. Anything that doesn't directly feed the agent strategy or enterprise revenue is vulnerable. For competitors, this creates openings. If you're building in spaces OpenAI is abandoning, you just got breathing room. For enterprises evaluating AI vendors, this is actually good news. A focused OpenAI with clear priorities is more reliable than one throwing products at the wall. And for anyone building AI products, here's the lesson: brand matters more as you scale, and some revenue isn't worth the reputational cost.

Source: The Verge AI