Summary: In its March 10 report, The Information drew on sources with knowledge of the effort and said OpenAI intends to launch Sora’s AI video generation capabilities inside ChatGPT “soon,” framing it as a strategy shift rather than the routine addition of one more feature. The go-to-market story gets simpler too: one destination to market instead of two.
Sora 1’s run in the United States came to a formal end on March 13, 2026. OpenAI had made the schedule public on February 27, yet the cutoff still caught people off guard: from that day forward, American users are left with a single video generation tool, Sora 2, and every image and video that stayed unexported inside Sora 1 was wiped. That very same week, The Information reported that OpenAI intends to move Sora’s video generation capabilities over to ChatGPT. Set the two events side by side and the message could hardly be clearer: in the roadmap OpenAI is working from, video generation is no longer a product of its own — it is one more feature living inside the chat interface.
Deadline Day: March 13
Announced on February 27, the cutoff gave US users roughly two weeks to act, and the version that circulated through the community forum pulled no punches: “All images and videos will be removed,” paired with the official instruction to “please download any assets you would like to keep.” Whatever the label, the substance was a product being shut down, not upgraded. Any US user who skipped a manual export before March 13 lost their work outright, and nothing on the far side of that date offers a way to get it back. Sora 2 is the sole approved landing spot for US users going forward, yet no automatic migration links the two versions; content, projects, and workflows all have to be rebuilt by hand. The geography deserves a note as well: the United States went first, a detail worth holding onto when similar notices for other regions arrive on their own separate clocks.
Sora’s Video Engine Heads Into ChatGPT
In its March 10 report, The Information drew on sources with knowledge of the effort and said OpenAI intends to launch Sora’s AI video generation capabilities inside ChatGPT “soon,” framing it as a strategy shift rather than the routine addition of one more feature. Mashable’s follow-up added the context that this consolidation is being planned a mere few months after the Sora standalone app went live — a remarkably quick about-face for an app presented as a destination in its own right. Line the events up in order and the trajectory tells itself: Sora 1 retires in the US first, then word spreads that video is being folded into ChatGPT; as one door closes, the remaining surface absorbs its function, and the product line tightens around a single entry point. Timing and packaging have not been confirmed by anyone official, so the rollout specifics remain unknown — but as signals go, this one is impossible to miss.
The Data Lesson for Creators
For anyone who runs real work through generative AI tools, this shutdown doubles as a warning. To begin with, the default is that nothing stored on an AI service belongs to you: revise the server-side deletion policy once, and every unexported item is simply gone — no channel exists to get it back, and no appeal process either. Next, swapping versions is not the same as migrating: the step from Sora 1 to Sora 2 shipped with no built-in transfer, which left even the diligent rebuilding their workflows by hand while an arbitrary date on the calendar bore down. Third, spotting the danger was entirely your job — the notice lived in help documentation and community channels, not in a modal that interrupts every login. The habits that actually help are uncomplicated: move every generated asset into storage you control the moment it lands, treat the AI tool as a machine on the production line rather than the filing cabinet where finished work is kept, and maintain a running export of anything you would genuinely hate to lose. Teams wired in through APIs should track product notices all the more closely, because a sunset date reaches you ahead of time but the calendar never moves to suit your readiness.
The Commercial Math Behind the Merge
The reasoning behind tucking Sora into ChatGPT does not take much untangling. A standalone app carries its own acquisition costs, its own retention metrics, and a constant case to make that the investment is still worth it; recast as a ChatGPT feature, video generation simply rides on a subscription that already exists — the single subscription reads as worth more, and user attention never leaves the one interface. The go-to-market story gets simpler too: one destination to market instead of two. This is the everything-under-one-roof playbook in action — capabilities keep multiplying at the model layer while the product surface narrows to a single door. Read across the competitive field, the coming fight over video generation is unlikely to be settled by shipping yet another standalone video app; the winner will be whichever primary interface can absorb more capabilities without losing focus, a question of interface design at least as much as one of model capability. Developers, meanwhile, should watch for the moment and the shape of video generation’s return across the ChatGPT API and product surfaces — and for what happens to pricing once video stops being a product with its own line item and becomes a checkbox inside an existing plan.