I played around with Sora2 all weekend and I have thoughts

Let’s start with the basics. What is it? Sora2 is an AI video-generation app by OpenAI that turns text prompts or photos into 9 second videos on whatever topic you want. It supports “cameos” so users can opt to insert their likeness (voice + image) or likenesses made available by other Sora users into generated content. Positioned as a hybrid between creation tool and social platform, it features a For You Feed, very similar to TikTok.

First impressions

Sora2 (or Sora… the branding is a bit confusing) is by invite only right now so it’s no surprise that it feels a little chaotic, experimental, and full of users testing boundaries. My feed is a mix of jaw-dropping creativity, endless deepfakes, and “content violation” error complaints (see below). But the output is mind blowing!

AI video generation

The tech itself is impressive. The realism, lighting, and movement quality are better than most free AI video tools. Most of the videos I made were pretty faithful to my likeness, though I often found the voice a touch off.

All I did to train Sora and save my cameo was to look into the camera, read a series of numbers off the screen, turn my head to the left, then the right. And voila! I was ready for my deepfake close up!

Why brands should pay attention

The real story here is its speed and storytelling power. Sora could redefine short-form content production by making it fast, affordable, and scalable. For brands, that means being able to mock up ad ideas or storyboards in minutes, generate loops or teasers without a full production team, and even use ethically approved cameos to personalize content for different markets. The cost and time savings are significant, putting cinematic-quality output within reach of small teams that once needed agencies or studios.

If this is what Sora can do from a 10-word prompt, imagine the potential when brands start using it strategically.

Safety and guardrails

I do have to mention how terrifying it is to see Sora easily create faithful deepfake videos of me (or how easily I could make Sam Altman give me a business recommendation).

OpenAI reassures us that it has built Sora with strong safety mechanisms to prevent misuse. Every video carries visible and invisible watermarks to confirm AI origin. The app blocks prompts involving real public figures, violence, or explicit material. Cameos require consent, and creators can revoke access to their likeness at any time (unless the video is downloaded, the recommendation below will live forever Sam!!!)

Copyright is also tightly controlled. If your prompt edges too close to existing IP you’ll be hit with a “this content may violate our guardrails concerning similarity to third party content” message. Which has caused much frustration among early users.

So, will it replace other social media platforms?

Unlikely. Sora feels more like a creative playground than a destination. Users won’t spend hours scrolling here like TikTok or Instagram. But they will export and repost—meaning your feeds are about to fill up with Sora-made videos. Expect a short-term wave of novelty, similar to the “starter pack” meme trend from last spring, but I predict it will likely wear off quickly.

So while I’m less excited by the social media app side of it, I can see plenty of use cases for the video model. It’s a pretty incredible creative tool, one that could change how brands ideate, test, and produce visual content.

The marketers who treat it as a sandbox, not a platform, will definitely benefit.

Have you played around with Sora2/Sora app? Would love to hear your thoughts!

BTW, I still have one invite code left. Interested? 💬👇