AI Stopped Being a Side Project This Week
OpenAI shutdown its flashiest product. Shopify made your store visible inside ChatGPT. And Anthropic accidentally leaked its most powerful model. One pattern connects all three.
Something important happened this week, and it wasn’t any single announcement. It was the pattern.
On Monday, OpenAI shut down Sora, the AI video app it launched six months ago with a billion-dollar Disney partnership attached. The same day, Shopify flipped a switch that made millions of stores discoverable inside ChatGPT. And by Thursday, Anthropic had accidentally leaked internal documents about a model it’s calling the most capable it has ever built.
Three stories, three companies, one direction: AI is done being experimental. The companies building it are cutting anything that doesn’t drive real business value and doubling down on the infrastructure that does.
If you run a service business, this week told you where the next two years are going.
1. OpenAI Shut Down Sora and Walked Away From a Billion-Dollar Disney Deal
Six months ago, OpenAI launched Sora as the future of AI video. It was a TikTok-style app where users could generate videos with AI. Disney signed a billion-dollar deal to license characters for it. It shot to the top of the App Store.
This week, OpenAI pulled the plug.
The numbers tell the story. Sora was costing roughly a million dollars a day to run. Downloads peaked in November and then fell off a cliff, dropping about 66% by February. Total lifetime revenue from in-app purchases came to $2.1 million. Against daily operating costs in the millions, the math was never going to work.
But it wasn’t just economics. Sora had become a liability. Users were generating deepfakes of public figures despite moderation rules. Robin Williams’ and Martin Luther King Jr.’s daughters both had to publicly ask people to stop making AI videos of their deceased fathers. Copyright issues were constant. The moderation burden was enormous.
The most telling detail: Disney, which had committed a billion dollars to the partnership and was planning a Disney+ integration, found out about the shutdown less than an hour before the public announcement.
This is OpenAI making a decisive strategic choice. Fidji Simo, who now runs day-to-day operations, held an all-hands meeting earlier this month telling employees the company was done with “side quests.” Everything is being consolidated around ChatGPT and enterprise tools. Consumer experiments that burn compute without generating revenue are over.
OpenAI’s head of Sora admitted publicly that “the economics are completely unsustainable.” That kind of candor is rare from a company preparing for an IPO.
2. Shopify Made Every Store Discoverable Inside ChatGPT
While OpenAI was shutting down its flashy consumer product, something much more consequential launched quietly on the same day.
Shopify activated Agentic Storefronts for all eligible merchants. What that means in plain terms: when someone asks ChatGPT for a product recommendation, Shopify stores can now appear in the conversation. No app to build. No integration to configure. No ad budget required. It is on by default.
This is not a pilot with a handful of big brands. This is millions of Shopify merchants, discoverable through conversational AI, managed from the same Shopify dashboard they already use.
The system also connects to Google’s AI Mode, Microsoft Copilot, and the Gemini app. Shopify is positioning itself as the commerce infrastructure layer for every major AI platform simultaneously.
The numbers underneath this are significant. AI-driven traffic to Shopify stores is up 7x since January 2025. AI-attributed orders are up 11x over the same period. Those are not theoretical projections. That is actual transaction data.
Here is the part that matters most for business owners: this is changing how customers discover products. Instead of searching Google, clicking through results, and landing on a website, a growing number of people are asking an AI assistant. The AI makes a recommendation. The customer clicks through and buys.
If your products or services are not structured in a way that AI can understand and recommend, you are invisible in this new channel. That is true whether you sell products on Shopify or run a service business that people find through search.
Discovery is shifting from search engines to AI conversations. This week made that concrete.
3. Anthropic Accidentally Leaked Its Most Powerful Model
On Thursday, cybersecurity researchers discovered that Anthropic had left nearly 3,000 internal documents in a publicly searchable database due to a configuration error in its content management system.
Among those documents: a draft blog post describing a new model called Claude Mythos that Anthropic says is “by far the most powerful AI model we’ve ever developed.”
The model sits in a new tier called Capybara, above the existing Opus line that was previously Anthropic’s most capable offering. According to the leaked draft, Mythos scores dramatically higher than Claude Opus 4.6 on tests of software coding, academic reasoning, and cybersecurity.
Anthropic confirmed the model exists after Fortune broke the story, calling it “a step change” and “the most capable we’ve built to date.” They also confirmed they are testing it with a small group of early-access customers focused on cyber defense.
Two things stand out. First, the capability jump is reportedly large enough that Anthropic created an entirely new pricing tier for it, not just an update to the existing lineup. Second, Anthropic is being deliberately slow about releasing it, citing cybersecurity risks. The leaked draft warned that the model “poses unprecedented cybersecurity risks” and could enable attacks that outpace defenders.
The irony of a company warning about cybersecurity risks while accidentally leaving its most sensitive documents in an unsecured public database is not lost on anyone. But the underlying signal is real: the next generation of AI models is materially more capable, and the companies building them are starting to slow down releases because the power demands more caution.
For business owners, the practical implication is straightforward. The AI tools available to your business six months from now will be significantly more capable than what is available today. Building the organizational muscle to use AI well -- clean data, documented processes, clear governance -- matters more than picking the perfect tool right now.
What This Means
Pull back from the individual stories and the pattern is clear.
OpenAI shut down a flashy consumer product that burned money and doubled down on enterprise infrastructure. Shopify launched a system that makes commerce work natively inside AI conversations. Anthropic is sitting on a major capability jump and being cautious about when and how to release it.
AI is growing up. The “let’s launch everything and see what sticks” era is over. What is replacing it is focused, infrastructure-level integration into how businesses actually operate: how customers find you, how they buy from you, how work gets done.
This is not a future prediction. This is what happened in a single week.
What Business Owners Should Actually Do
If you sell anything online, check whether your products are showing up in AI conversations. The Shopify announcement is the most visible version of this, but the principle applies broadly. AI assistants are becoming a discovery channel. If your business is not structured for AI to understand and recommend, you are missing a channel that is growing fast.
Stop waiting for AI to “settle down” before engaging. OpenAI just shut down a product that was six months old with a billion-dollar partnership behind it. The landscape will keep shifting. The advantage goes to businesses that are building capability now -- learning what works, training their teams, automating where it makes sense -- not ones waiting for the dust to settle.
Focus on your data and processes, not on picking the right AI vendor. Anthropic just revealed a model tier that did not exist last week. OpenAI is consolidating its entire product line. Google and Microsoft are both racing to embed AI into their platforms. The vendor landscape will keep changing. What will not change is that the businesses with clean data, documented workflows, and a culture that can adapt to new tools will be the ones that benefit most from every new release.
Pay attention to how your customers are finding you. The shift from traditional search to AI-assisted discovery is not theoretical anymore. It showed up in Shopify’s transaction data this week. If your business depends on being found online, understanding how AI recommends and surfaces businesses is becoming as important as SEO was ten years ago.
The Bigger Picture
The week of March 24, 2026 will probably not make the history books. No single story this week was the kind of headline that stops you in your tracks.
But taken together, these three developments mark something real. The biggest AI companies in the world are making hard choices about what matters, and what they are choosing is business infrastructure over consumer spectacle. Integration over experimentation. Revenue over demos.
For business owners, the signal is straightforward. AI is becoming part of how commerce, customer discovery, and enterprise operations work at a fundamental level. Not as an add-on. Not as an experiment. As infrastructure.
The businesses that start treating it that way now will have a meaningful head start. Not because they picked the right tool, but because they built the organizational readiness to use whatever tool comes next.
That readiness -- clean data, clear processes, a team that knows how to work with AI -- is the real competitive advantage. And it is one you can start building today.