In September 2025, Disney sent Character.AI a legal demand over its characters showing up in AI chatbots. In September 2026, Disney gave the company’s former CEO its first-ever CTO job. Same company, same characters, twelve months apart.
Karandeep Anand, who ran Character.AI, now oversees technology, data, and AI platforms at Disney. That’s the whole story, and it’s enough to make anyone who evaluates AI tools for a living stop and recalculate a few assumptions.
One housekeeping note before the analysis: some of the aggregated coverage floating around attributes this hire to Jamie Brew, a Character.AI co-founder. The primary reporting names Karandeep Anand, the former CEO. If you saw the other version in a feed somewhere, that’s a secondhand summary drifting from the source. I mention it because the speed at which AI-adjacent news gets garbled is itself part of the story.
What this actually tells us
I spend most of my time testing tools and writing up what holds together and what falls apart under load. The recurring problem in this category isn’t capability. It’s provenance. Where did the training data come from, who owns the outputs, and what happens when a rights holder with real lawyers decides to care.
Disney has been the most aggressive rights holder in that conversation. It went after Midjourney. It went after Character.AI. Its position read as absolute: our characters, our property, don’t touch them. That posture gave a lot of us a convenient shorthand when reviewing tools. If a product generated recognizable IP, it carried risk, and Disney was the reason why.
Hiring the CEO of a company you accused of infringement complicates the shorthand. It doesn’t reverse the lawsuit and it doesn’t mean Disney has softened on enforcement. What it suggests is that Disney separates two questions that most commentary mashes together: is this person building something we object to, and is this person good at building AI systems at scale. Disney apparently answered yes to the first and yes to the second, then hired on the second.
The practical read for people picking tools
A few things I’d adjust in how I think about this category:
- Legal posture is not a product review. A company suing over IP today may be recruiting from the defendant’s org chart tomorrow. If your tool selection depends on a rights holder’s mood, your tool selection is unstable. Evaluate the license terms and the indemnification language, not the headlines.
- Enforcement and adoption run on parallel tracks. Large media companies are simultaneously litigating against AI firms and building internal AI capability. Treat those as separate programs with separate budgets, because that’s what they are.
- Talent moves faster than policy. The number of people who have shipped consumer AI products at scale is small. That scarcity gives them mobility that cuts across legal lines, and it means the people who built the tools you’re testing may end up governing the platforms you test them against.
- Character-based AI is not going away. Character.AI’s core idea, conversational personas people talk to for hours, is exactly the mechanic a company with the deepest character library in entertainment would want to control rather than ban.
That last point is the one I’d bet on. Disney didn’t hire a general infrastructure executive. It hired someone whose previous job was running a platform built around persona-driven conversation, and it put him in charge of technology, data, and AI platforms. You don’t need a leaked roadmap to see the shape of that.
What I’ll be watching in the product
The interesting question for reviewers isn’t the hire, it’s what ships. If Disney builds official character experiences, the comparison set becomes obvious: how does a first-party, licensed, rights-cleared character chatbot stack up against the unlicensed versions people already use. That’s a test worth running, and it’s the first version of that test where the licensing question has a clean answer.
My guess, and it is a guess, is that the first-party version will be safer, more restricted, and less fun, at least initially. Guardrails on a corporate-owned character are heavier than guardrails on a user-generated one, and the legal exposure of an official Mickey saying something off-script is enormous. The tradeoff between fidelity and freedom is the thing I’d want to measure.
The honest summary
A company sued an AI startup, then hired its CEO to run its technology stack. Both things are true, neither cancels the other, and anyone treating this as hypocrisy is missing the more useful reading: Disney wants the capability and intends to own it rather than license it from someone who built it on borrowed material.
For those of us testing these tools, the takeaway is smaller and more practical. Judge products on terms, output quality, and what you’re actually allowed to ship. Corporate legal positions are weather, not climate.
🕒 Published: