# OpenAI killed the tool behind this AI film. Its director says Hollywood is only getting started > Source: > Published: 2026-07-20 09:04:00+00:00 *Critterz* was one of the first feature films built primarily with [AI](https://www.fastcompany.com/section/artificial-intelligence). But OpenAI abruptly shut down Sora, the tool its director, Nik Kleverov, was relying on. Kleverov, who is chief creative officer of the production company Native Foreign, makes the case that AI isn’t killing Hollywood but is about to reinvent it. He breaks down why 2026 is year zero for mass adoption of AI in the film industry, and why human talent is more in demand than ever when great storytelling is the goal. *This is an abridged transcript of an interview from* [Rapid Response](https://mastersofscale.com/episode_category/rapid-response/)*, hosted by former *Fast Company *editor-in-chief Robert Safian. From the team behind the *Masters of Scale* podcast, *Rapid Response* features candid conversations with today’s top business leaders navigating real-time challenges. Subscribe to *Rapid Response* wherever you get your podcasts to ensure you never miss an episode.* *Critterz* was initially slated to premiere at the Cannes Film Festival this year. Yeah, right about now. **But it got derailed a little bit by technical issues. Or was it that the Cannes film folks don’t really like AI?** Interesting you should mention that, because the answer is, of course, that the core technology we were building the film with went away in the middle of making the film. But I do remember when we made the announcement late last fall that, hey, OpenAI is backing our film, so on and so forth. There was almost a pre-boycott. I was reading through it. There were French groups and people saying, “No, we will not have this at Cannes.” It was like, well, yeah, OK. We were obviously also thinking you can’t guarantee that you’re going to play at the Cannes Film Festival. So it was more of an OpenAI endeavor that we were embarking on. **But you were able to get in there.** Sure, yeah. **The initial film idea came from a short that came out of OpenAI, right?** That’s right. In the Stone Age of 2022 of AI video, shortly after DALL-E came out, which is OpenAI’s text-to-image platform, we actually pitched them a short film idea. After my partner, Chad Nelson, and I created this world of *Critterz*, we pitched them this idea of, “OK, well, what if we did a float-through of this beautiful forest with all these critters?” So you get that tech demo part of it handled. It starts as a bit of a David Attenborough Planet Earth, and you’re hearing this omniscient narrator talk about this majestic world. But then, once we meet the critters and see them in their natural environment, it becomes a Monty Python sketch, where it’s like, no, you’re all wrong. That’s actually not what we’re doing here. It’s just really irreverent, silly bickering back and forth between the characters on screen and this voiceover. And that was it. It was a super simple 5-minute short. It ended up being much more successful than we had ever imagined. I think we pitched it in the winter of ’22, then made it over holiday break, and released it for the one-year anniversary of DALL-E, which was April of ’23. People really liked it and were curious about the process. Remember, back then, yes, it was generated with AI, but all the video was traditional. It was Unreal Engine, it was After Effects, and it was all these tools that we all know in the industry and use. So really, the generative part of it was the creation of these characters and the world. Everything else was the way that we do creative. **And when you moved to making it feature length, you started using more generative tools. That’s when you started using Sora?** Well, yeah, because if you look at the timeline, in ’23 we released it. Sora was announced in spring of ’24. So between ’23 and ’24, the *Critterz* film actually created a good relationship between Native Foreign, my company, and OpenAI. We were doing creative work for them, agency work. Then they were like, “Hey, we’re going to go multimodal. Are you interested in testing this out?” So I was playing with Sora before it had a name, and I was one of the initial—they call us the Sora Seven, as if we went to the moon, which is silly. **When Sora got pulled, which is what slowed your roll a little bit with Critterz, were you surprised? Did you get early insight on that the same way you got early insight on it coming out?** I’d love to say I was that cool, but no. Unfortunately, we found out around the time everyone else did. It was an interesting day, because being in the middle of something—and of course, having always been big fans of OpenAI and being tied to that tech—was an interesting challenge for us as filmmakers. I think it’s a great case study for where we are in this world, because here we are pushing one of the first features, especially animated and four-quadrant-type projects, forward, and then boop, our main technology goes away or becomes obsolete. Granted, they were keeping the API open for us and everything. But at the same time, we have to really think about the process and say, “OK, well, does it make sense to use something that’s not going to be supported several months down the line? Or do we take some of the calls?” Because, of course, everyone did call that day. There were plenty of companies that wanted to work on it with us, and we were, at the time, for that project only, doing an exclusive. That’s where we’re at now. So we partnered with AGC [Studios], and we brought the film to Cannes Film Festival, not necessarily to sell it, but actually just to start the conversations. **So it’s not the full film. It’s pieces of the film that you could bring.** Pieces of the film that we brought and were showing. **And those pieces were done on Sora? Or were they redone using some of the new other tools?** Both. So there was a lot on Sora. And then, in the time between March and May, we said, “OK, let’s start rebuilding and doing some new camera tests.” We call them camera tests even though they’re model tests. It’s like, “OK, what does this lens look like? What does that lens look like?” But by lenses, I mean the generative programs. **And each one is different.** Of course. **And the way you use it is different.** The crazy thing is 10 different people can sit down with the same tool and get different results. So when people say, “Humans aren’t important anymore, the tools are all the same,” it’s absolutely untrue. Depending on which creative is at the steering wheel, you get very different outputs. **Just as having the same camera doesn’t mean you end up with the same film.** Look, there’s a lot of talk about democratization and all that, but I will also say everyone’s got a camera in their pocket. And how many 13-year-olds are making the next masterpiece film? There’s some truth to the idea that there will be undiscovered talent from all over the world that finally has a shot and is going to do a lot of stuff. But I’m very much not of the Hollywood-is-dead narrative, because storytelling is hard, and making really great storytelling that transcends is still hard. **I find with other AI that I use sometimes, because it’s so easy, seemingly, to get more variation, I can go down a rabbit hole and keep trying to create new things. So I’m curious how that works for you, because you can think, “Oh, let’s try this one. What if we did it this way?” And that can expand.** You have to have restraint. I think this is where creative direction and vision come in really critically, because you can meander forever. As the tools have gotten better, the ratio has gotten better. Some shots I literally made 1,000 generations before I got to one that was decent enough that I still had to finish out in VFX. Now we’re maybe 5-to-1, 10-to-1, depending on whether we’ve established a really strong key frame of what we’re able to do. So things have gotten much more efficient. But yeah, there’s a serendipity, and you have to be malleable enough, but you also have to keep it on the rails. That’s an interesting balance. **It’s only been a few years that we’ve had these capabilities in this way. How far along are we? How fast is it changing?** In many ways, and you talked about Hollywood studios earlier, we’ve been having discussions for a long time, but 2026 really seems like year zero, where people are saying, “OK, we’re going to try some stuff out now. We’re going to actually move forward and make things happen.” So I think we’re just at the beginning. There’s been a lot of sorting out that’s had to happen, and it’s go time.