Disney hired the CEO it once sent a cease and desist to run its AI

Three moves landed in a week, and none of them were about a new model. Disney named its first CTO. Anthropic handed an outside firm the keys to watch how its models get built. Meta put a price tag on its AI tools. Different companies, same signal: the AI bet is now showing up in org charts and billing pages, not demo reels. Let me catch you up.

The deep cut

• AI strategy now lives in hiring and pricing. Disney, Anthropic, and Meta each moved AI out of the lab and into the org chart.
• Hire the person who understands the threat. Disney picked Karandeep Anand, the Character.AI CEO it once sent a cease and desist.
• Charge for the AI you already spent on. Meta turned a $14.3 billion Scale AI deal into Core and Premium subscriptions.

Hiring your way to an AI answer

Disney gave itself a chief technology officer for the first time, and the pick says plenty. Karandeep Anand ran Character.AI, the startup Disney hit with a cease and desist last year over copyrighted characters. Now he reports straight to new CEO Josh D'Amaro and owns infrastructure, product, engineering, and AI platforms.

Read it as a bet on knowledge over clean history. Anand spent years at Microsoft and Facebook before the startup, and he knows exactly how generative tools chew through IP because he ran a company accused of doing it. The hire also comes as Disney's OpenAI partnership fell apart, so Disney wants someone who can build in-house, not rent.

Letting an outsider into the training room

Anthropic did the opposite of guarding its process. It brought Accenture inside to evaluate and red-team its models, with both sides committing at least $1 billion each over five years. The work runs through Faculty, Accenture's AI arm, and these evaluators get access "comparable to an employee's." They watch models take shape in training and talk to staff directly.

This is new ground, and Anthropic admits it. There are no standards yet for what embedded evaluators see or how they report. Anthropic says independent eyes "do not reduce our accountability, but help to make it more verifiable." The honest part: they are funding Accenture directly for now because no shared or government funding pool exists. Trust, with an invoice attached.

The bill comes due

Meta stopped treating AI as a free perk. Meta One bundles app subscriptions with extra AI usage, starting at $7.99 for Core and $19.99 for Premium, with business tiers climbing to $499 a month. The point is monetizing the Muse models that grew out of its $14.3 billion Scale AI investment. BNP Paribas pegs the subscription push at $13.5 billion in revenue by 2028.

The early paid tiers already move numbers. Instagram's daily revenue jumped 475% to $1.2 million the week the plans took hold. For your team, the detail worth noting is how Meta framed paid visibility. Asked whether this hurts creators who cannot pay, Meta's Helen Ma said the company stays "very committed to keeping the experience for creators free." Journalists in the room still asked if paying to get noticed becomes a cost of doing business.

The UI stops being the moat

Underneath these moves is a shift in what companies defend. Salesforce, once loud about AgentForce, is now leaning into Anthropic and OpenAI chatbots as the preferred interface and charging a premium for it. The read from Stratechery: Salesforce is giving up UI as a moat "because it's disappearing for everyone."

Anthropic is building in the same direction. Claude Code relaunched Projects to run multiple agents under one roof, each thread a cloud session with a coordinator sorting the overlap like a merge conflict. The interface matters less than the engine behind it. If your product's edge is a screen people love, check whether an agent could do the same job without it.

Three questions for your team

• If we had to name one person to own AI tomorrow, do we want deep domain knowledge even if their past clashed with ours, the way Disney picked Anand?
• Where does our AI spend turn into revenue, and can we point to a line item the way Meta points to Muse and Meta One?
• If the interface stops being our moat, what are we actually defending, and would we let an outside evaluator look under the hood to prove it works?