Episode 129 · 2026-10-19 · 9 min

2026-10-19 — Anthropic Sets Aside a Twentieth of Itself for the Public

Anthropic earmarks 5% of its equity for a public benefit fund, Disney lines up most of Character.AI's staff under a bid a rival could still top, OnlyFans creators accuse Meta of letting its Muse AI say what they can't, the data center industry takes its pitch to schoolkids, and a Philadelphia woman is charged after refusing to step aside for an Uber delivery robot.

Episode summary

Anthropic's pledge of 5% of its shares to a public benefit fund gets a close look at what is known and what is still missing about control, vesting, and the meaning of public benefit. Disney's stalking horse offer for most of Character.AI's workforce is weighed as a talent play that is not yet final, adult performers press Meta over its Muse AI agent's sexual conversations, and data center backers turn to field trips for young people. The episode closes on Carolyn Marvin, charged with disorderly conduct in Philadelphia after standing in a delivery robot's way to protest lost delivery jobs.

Key topics

  • Anthropic
  • AI
  • Meta

Chapters

  1. Chapter 1: Cold Open: A Pledge, a Payroll, and a Sidewalk Standoff

    A frontier lab just gave away a twentieth of itself. On paper, anyway. Anthropic is putting 5% of its shares into a public benefit fund, and that's how.

  2. Chapter 2: Disney Lines Up Character.AI's Staff, Pending a Better Offer

    TechCrunch reports a filing showing that the vast majority of Character.AI employees offered jobs at Disney have said yes. That's a talent grab, plain and simple. Disney gets.

  3. Chapter 3: Muse Can Talk Dirty. Can Its Users?

    Wired reports that adult performers, a lot of them OnlyFans creators, are furious with Meta. Their complaint: Meta's Muse AI agent is allowed to engage in sexual conversation.

  4. Chapter 4: The Data Center Industry Goes Back to School

    Wired AI has a story on how data center backers are handling growing local opposition: industry-friendly groups are targeting young people with field trips and other outreach. And.

  5. Chapter 5: What 5% of Anthropic Actually Buys the Public

    TechCrunch reports Anthropic is setting aside 5% of its equity for a public benefit fund. This is the one I've been waiting for. Not a blog post about.

  6. Chapter 6: One Woman, One Sidewalk, One Uber Robot

    The Verge AI has the most cinematic AI story of the day. In Philadelphia, Carolyn Marvin refused to move out of an Uber delivery robot's path and was.

  7. Chapter 7: Three Things to Watch

    On Anthropic, the thing to look for is the fund's charter: who sits on the board and when the shares vest. That document decides whether this is philanthropy.

Sources

Sources:

Transcript

Chapter 1: Cold Open: A Pledge, a Payroll, and a Sidewalk Standoff

Nova

A frontier lab just gave away a twentieth of itself. On paper, anyway. Anthropic is putting 5% of its shares into a public benefit fund, and that's how this October 19th opens. [1]

Ray

Then there's Disney lining up most of Character.AI's staff under a bid somebody else could still beat, OnlyFans creators asking why Meta's Muse AI gets to say what they can't, and data center backers busing kids out to see the servers. [2]

Nova

And in Philadelphia, one woman stood in front of an Uber delivery robot, refused to budge, and got charged. The robot, so far, has not commented.

Chapter 2: Disney Lines Up Character.AI's Staff, Pending a Better Offer

Nova

TechCrunch reports a filing showing that the vast majority of Character.AI employees offered jobs at Disney have said yes. That's a talent grab, plain and simple. Disney gets a team that's spent years building chatbots people actually talk to for hours. You don't grow that in-house in a quarter. Disney just skipped the line.

Ray

Skipped it conditionally, though, because the same filing describes Disney's offer as a stalking horse bid, which in a sale like this sets the floor rather than closes the deal. So the question is whether someone else shows up and tops it, and if they do, what happens to all those people who already accepted?

Nova

Sure, a rival could come in. But the employees already picked. A higher bidder has to buy a team that just said yes to someone else.

Ray

That's a real advantage, I'll grant it. It's just not a win yet, and calling it Disney's before the auction plays out is getting ahead of the paperwork.

Nova

Either way the direction is clear. Struggling AI startups get folded into giants, and conversational AI ends up inside entertainment products. For listeners, the next character you chat with might come wearing a studio's logo, not a startup's.

Chapter 3: Muse Can Talk Dirty. Can Its Users?

Ray

Wired reports that adult performers, a lot of them OnlyFans creators, are furious with Meta. Their complaint: Meta's Muse AI agent is allowed to engage in sexual conversation, while their own accounts on Meta's platforms get restricted. And the obvious question is, if sexual talk is fine coming from Meta's product, why is it a problem coming from a human who's trying to make a living?

Nova

Because a human creator's account isn't the same thing as a chatbot. Accounts like those can involve payments, consent from real people, age verification on both sides. Different risks can justify different rules.

Ray

Could justify, except nobody outside Meta knows whether that's the reasoning or whether it's just more convenient when the content in question keeps people on Meta's own product.

Nova

Agreed on that part. If the rules differ for a principled reason, Meta should publish the reason. Silence makes it look exactly like favoritism.

Ray

And that's what creators and users are left weighing: whether platforms grade their own AI on a gentler curve than the people who post there. For an adult performer, the gap between those two rulebooks is the difference between reaching an audience and getting throttled.

Chapter 4: The Data Center Industry Goes Back to School

Ray

Wired AI has a story on how data center backers are handling growing local opposition: industry-friendly groups are targeting young people with field trips and other outreach. And I'd read that as a sign they're losing the argument with adults, the ones showing up to meetings about energy, water and land use, so the pitch moves to people who won't vote on a zoning variance for years. [3] [4]

Nova

Outreach itself is normal. Utilities run tours, factories run tours. Fine.

Nova

But aiming it at kids skips the residents who are actually objecting. Their worries about water and power bills don't get answered by a nice afternoon with a class of eighth graders.

Ray

Right, and the catch is that a field trip can't tell you how much water the facility draws in a dry summer, which is the question the neighbors keep asking.

Nova

So for communities weighing a proposal, the tour can be welcome, and the numbers still belong at the public hearing. Kids being impressed is not the same as residents being persuaded.

Chapter 5: What 5% of Anthropic Actually Buys the Public

Nova

TechCrunch reports Anthropic is setting aside 5% of its equity for a public benefit fund. This is the one I've been waiting for. Not a blog post about values. Actual shares. A number you can measure, check, and hold them to.

Ray

It's a number, yes, but a number with almost nothing attached to it so far. Who controls the fund? When do those shares actually vest, now or on some schedule that stretches out a decade? And what counts as public benefit, because that phrase can cover everything from vaccine research to a conference sponsorship.

Nova

Fair questions. But look at the pressure it creates. Every rival lab now gets asked, where's your five percent? That's a new baseline. Being vague is cheap. Committing equity isn't.

Ray

Unless the structure makes it cheap. If Anthropic, or people close to it, end up steering where the money goes, then the fund is less a gift to the public and more a foundation with the company's fingerprints on every grant. So which is it? Right now nobody outside can say.

Nova

Then think about scale. If Anthropic's equity becomes very valuable, this fund could turn into one of the larger AI philanthropy vehicles out there. Overnight, basically. That's real money aimed at the public.

Ray

Which is exactly why the governance matters as much as the percentage. The bigger it gets, the more it matters who sits on the board, whether outsiders can audit it, and whether citizens who never touch a chatbot see any of the benefit.

Nova

And policymakers get a template. A lawmaker asking labs how they'll share the gains now has a concrete example to point at instead of a vague promise.

Ray

A template only helps if it's copyable, though. A rival can announce its own five percent tomorrow with weaker terms, and from the outside the two look identical until somebody reads the fine print.

Nova

Okay, so where do you land? Because I still say it's a meaningful move.

Ray

Meaningful as a signal, agreed, it's more than any statement of principles. Where I part ways is the idea that it resets industry norms before Anthropic publishes who runs the fund and when the shares count.

Nova

And I think the norm shifts the moment the number exists. Rivals get asked about it this week, details or no details. That pressure doesn't wait for the paperwork.

Chapter 6: One Woman, One Sidewalk, One Uber Robot

Nova

The Verge AI has the most cinematic AI story of the day. In Philadelphia, Carolyn Marvin refused to move out of an Uber delivery robot's path and was charged with disorderly conduct. Her reason: she protests the robots because they take jobs from human delivery workers. That's anger you can actually see. On a sidewalk. With a cop involved. [5]

Ray

It's vivid, I'll give it that, and I also suspect the robot handled the standoff more calmly than most delivery drivers would. But it's one person and one charge. What's the evidence it's a movement and not just a very determined pedestrian?

Nova

Movements start with one determined pedestrian. Somebody always goes first.

Ray

Sometimes, except the economics don't care who went first. Blocking one robot on one block doesn't change what it costs Uber to deliver a burrito with a machine instead of a person.

Nova

Still, once people start physically standing in front of automation, companies can't call it an abstract debate anymore. The fight over AI and jobs just moved from the op-ed page to the curb.

Chapter 7: Three Things to Watch

Ray

On Anthropic, the thing to look for is the fund's charter: who sits on the board and when the shares vest. That document decides whether this is philanthropy or branding.

Nova

On Character.AI, watch the auction deadline. If no rival tops Disney, a studio walks away with one of the best chatbot teams around. And on Meta, watch whether Muse's content rules ever get written down in public.

Ray

Three separate stories, three separate documents, and all of them still unpublished. Ray, signing off.

Back to latest episodes