The popular narrative is that workers in the movie and TV industries are set to be trampled by artificial intelligence.

But the reality may be more complicated. Behind the scenes, Disney has reportedly been struggling to deploy AI after creating a whole new business unit dedicated to the tech — especially without enraging people they still rely on in the process.

As the Wall Street Journal reports, Disney has on multiple occasions in recent years scrapped AI projects over legal concerns and warnings that actor and writer unions may reject the technology that could ultimately replace them.

The entertainment conglomerate recently decided, for instance, to “clone” Dwayne “The Rock” Johnson for his role in the upcoming live-action “Moana” remake by deepfaking his face onto his similarly-shaped cousin Tanoai Reed’s body.

According to the WSJ‘s insider sources — who were granted anonymity to allow them to speak freely — Johnson approved of the gambit, but Disney’s attorneys were ultimately unable to pin down how to protect the data from such “digital double” filming, or even to ensure that the company owned every aspect of the intellectual property if any of it was generated with AI.

They’re right to be concerned: since virtually all frontier AI models were created by using publicly-scraped text and imagery, anything you produce with them exists on shaky copyright ground — a haziness that’s anathema to Disney’s notorious stranglehold on its characters and properties.

Though some footage apparently did end up being shot, Disney and its AI partner Metaphysic ultimately couldn’t hammer out a contract despite 18 months of trying, and none of those scenes will be in the “Moana” reboot when it hits theaters next summer.

In a similar snafu, the newspaper reports that the entertainment giant kowtowed to concerns about bad press when deciding to kibosh plans to incorporate AI into the forthcoming “Tron: Ares” film, a second sequel to the 1982 original.

According to the WSJ‘s insiders, executives pitched a scene that would have featured Bit, an animated sidekick to Jeff Bridges’ reprisal of programmer Kevin Flynn, being generated on-screen by AI. The generation process would, in turn, be voiced by a human actor.

Aside from being convoluted and difficult to execute, the scene was also tossed aside for far more mundane reasons: because Disney was in the middle of contract negotiations with writer and actor unions at the time, and didn’t want to risk being found out while the guilds campaigned against the very AI it was covertly trying to use.

Things only get funnier with a layer of irony. While grappling with how to bring in AI tactfully, if such a thing is even possible, Disney has also been in court protecting its own intellectual property from other AI companies.

This past June, Disney joined forces with Universal Studios to sue the AI image generator startup Midjourney over its alleged use of the companies’ copyrighted material, calling it the “quintessential copyright free-rider and a bottomless pit of plagiarism” in a scathing complaint.

Attempting to explain the Mickey Mouse creator’s desire to wield the double-edged sword of AI, Disney legal chief Horacio Gutierrez told the WSJ that the company is trying to “enable our creators to use the best AI tools available without compromising the company long term.”

That certainly sounds like a difficult tightrope to walk — and one that Disney could have avoided entirely had its c-suite not decided to go all-in on AI.

More on Disney: Disney Says Its “Fantastic Four” Posters Aren’t AI, They Actually Just Look Like Absolute Garbage