Disney Just Hired an AI CEO as Its First CTO. Marketing Still Treats Talent Like Stock.

Last year Disney sent Character.AI a cease-and-desist for using Elsa, Moana, and Darth Vader without permission.

This week Disney hired the Character.AI CEO as its first companywide CTO.

Same 72 hours, an AI actress sat 75 simultaneous press interviews and started speaking Chinese mid-sentence.

That is not a cute entertainment recap. That is the latest AI news for anyone who still thinks synthetic talent is a Canva plugin.

What actually shipped

LAYER 1 — The brand put an AI operator next to the CEO

On 18 September 2026, Disney named Karandeep Anand Senior EVP and Chief Technology Officer. First time the company has had that role companywide. He starts 2 October. He reports to CEO Josh D'Amaro, not to a streaming sidecar.

Anand ran Character.AI for about 15 months after senior seats at Meta, Microsoft Azure, and Brex. A number of Character.AI engineers are expected to follow him. His brief covers data, AI platforms, product, engineering, and infrastructure. D'Amaro called the job “technology in service of creativity.”

NBCUniversal and Warner still keep tech chiefs inside streaming. Disney just put the AI product person in the room where the whole company gets decided.

LAYER 2 — Synthetic talent went on a junket and broke

Particle6 Group pushed Tilly Norwood, its AI “actress,” through a press tour for the film Misaligned. Up to 75 interviews at once. Paid public video chats after the first two minutes.

The same jokes recycled across CNN, Reuters, and the BBC. A capybara waiting room. Then, on camera with Piers Morgan and Tom Conti, the model switched into Chinese and had to apologize for crossed wires.

That is what happens when you scale a face before you lock a brief.

Marketing and AI: stop buying a face you cannot supervise

Here is the part most AI recaps skip, and it is the one that hits agency P&L first.

Big brands are hiring AI product people into the C-suite. Marketing teams are still buying AI influencers the way they used to buy stock photography. One prompt. One pretty still. No owner. No disclosure line. No kill switch when the avatar goes off-script on a live call.

We are not running it that way at Dzine Prodigy.

If a client wants synthetic talent, it is a production system, not a novelty hire.

LAYER 1 — Disclosure is the brief, not the afterthought. California already forced labels on AI-generated performers in ads. If you cannot say the person is not a person in one clean sentence, do not ship the asset.

LAYER 2 — One approved kit. One face. One voice. One prompt pack. Brand colors, banned claims, and a human who can pull the plug in under a minute.

LAYER 3 — Never give the model a live interview with no harness. 75 parallel junkets is not reach. It is 75 chances to say the wrong product name in the wrong language.

AI-native marketing in 2026 is not more avatars. It is fewer surfaces, tighter control, and a human on taste.

Do this before Friday

Pick one client. Write a one-page synthetic-talent rule: when we use it, how we label it, who signs off, who can stop it.

Then build one locked kit. Not a marketplace browse. One face that matches the brand bible.

Disney just admitted the operator layer belongs next to the CEO. Marketing does not get to keep treating talent like clip art.

The labs will keep shipping faces. Agencies that ship with a harness will still have clients next quarter.