OpenAI is currently one of the most powerful AI companies on the planet, and on March 24th, they proved it by doing three very human things at once: admitting a big idea didn't work, quietly killing a product that got too weird, and doing something genuinely good for kids online.
If that sounds like your typical Tuesday at work, congratulations; you might be running an AI empire.
First, the Funeral 🪦
Buckle up, because OpenAI just pulled the plug on Sora, its AI video app that was basically a deepfake factory in your pocket. Sora was supposed to be an AI-first TikTok, letting you clone your own face and star in AI-generated videos.
The Cringe Factor: People immediately started making deepfakes of Martin Luther King Jr. and Robin Williams, prompting their daughters to go on Instagram and beg users to stop. Yikes.
The Numbers: The app peaked at roughly 3.3 million downloads in November, then slid to about 1.1 million by February. Translation: the hype bubble popped faster than a viral TikTok trend. OpenAI says it's shutting Sora down to focus on other priorities, specifically redirecting its computing power toward robotics and world simulation research.
Oh, and remember that mega Disney deal? Disney had agreed to license Mickey Mouse and Cinderella for a planned $1 billion investment. That’s all dead now; Disney’s response was basically a very polite "good luck with that."
Then, the Stumble 🛒
While Sora was getting its eulogy, the "Instant Checkout" experiment was quietly limping home. Last September, OpenAI looked at Amazon and thought, "we could do that." They let users browse products from Walmart, Etsy, and Shopify without ever leaving ChatGPT.
The Reality Check: Users loved browsing, but when it came time to pay, they bolted back to websites they already trusted. Turns out, trusting a chatbot with your credit card is a bridge too far for most people; and honestly, fair!
OpenAI admitted the feature "did not offer the level of flexibility" they hoped for; so they are now letting merchants handle their own checkout experiences instead.
So here's where we are: one product dead, one product limping. You'd be forgiven for thinking this was a bad day at OpenAI HQ.
But here's the thing about OpenAI's chaotic Tuesday, it didn't end in a dumpster fire.
There’s in fact, a Win.
Underneath the shutdowns and the shopping flop was something actually worth celebrating. OpenAI released a set of open-source safety prompts that developers can drop directly into their apps to make them safer for teenagers.
The "Safety Starter Pack" covers:
Graphic violence and sexual contents.
Harmful body ideals and behaviours.
Dangerous activities and challenges.
Romantic or violent role play
Age-restricted goods and services.
OpenAI built these tools in partnership with Common Sense Media and everyone.ai. They called it a "meaningful safety floor", not a perfect solution, but a real step in the right direction.
So yeah, this is the full story of OpenAI’s Tuesday. A company big enough to fail publicly, pivot loudly, and still sneak in a genuine win all in the same news cycle. Love it or hate it, nobody is doing it quite like them right now.
