OpenAI has cut another product

This time it's "Adult Mode"—it was shelved before it even launched
But what's worth discussing isn't the "Adult Mode" itself, but what went wrong with OpenAI's product strategy.
Regarding Adult Mode: announced in October last year, originally scheduled to launch in December. As a result, all eight mental health advisors hired by OpenAI voted against it.
Reason: age verification system has a 12% failure rate—millions of minors could be exposed to adult content. Technically, it's also unfeasible: models were trained to avoid pornographic content, but now they need to generate content while filtering illegal material. The model can't handle this switch.
On March 26, it was indefinitely shelved.
TechCrunch's headline was very accurate: "OpenAI abandons yet another side quest." But that's not the main point.
The key pattern is: announcement → delay → shelve/cancel. What has OpenAI cut over the past year? Sora Video—secured a Disney deal, 3.3 million downloads, #1 on the App Store.
Shut down on March 24. Total revenue was $2.1 million, not enough to cover GPU electricity costs. Stargate expansion—failed to secure funding after talks with Oracle, canceled. In-app purchases—announced but canceled. Hardware devices—promised for this year, court documents show a delay until 2027.
See the pattern?
OpenAI's AI work is solid. GPT remains strong, the O-series has real substance. The problem is that every strategic expansion outside of AI has failed. Video, hardware, shopping, adult content—each direction is announcement, sprint, retreat.
Sam Altman said "AI + advertising is the last resort."
After 16 months, they finally launched ads.
When a company starts doing what it previously said it wouldn't, while cutting what it said it would—this is called strategic misalignment.