Disney and OpenAI Strike an Unexpected Deal - DTNS 5163  ## Document Notes ## Highlights - **Disney-OpenAI Partnership Is A Testbed** - OpenAI and Disney struck a three-year deal letting Sora and ChatGPT generate short fan-prompted videos and images using 200+ Disney characters and properties. - The partnership is framed as a "responsible entertainment AI" testbed with promised guardrails and age-appropriate controls. Transcript: Jason Howell So OpenAI and Disney just signed a three-year deal that lets Sora and ChatGPT image generation generate short, fan-prompted videos and images using more than 200 of Disney's characters, Costumes, vehicles, environments from all of its properties. And when we're talking about Disney, we're not just talking about Walt Disney movies and Disney TV and that sort of stuff. It is the Disney IP. It's also Pixar. It's Marvel. It's Star Wars. There's probably more in there. Apparently, all this kicks off in 2026. And this is really interesting to me. I want to see how this actually rolls out. But Disney Plus is going to offer a curated set of Sora clips from this partnership for its streaming subscribers. So when people go on to Sora and Sora, for those who might not know, is OpenAI's video generation model, does video, does audio generation tied to the video. It's actually pretty powerful. And so, you know, people creating stuff on Disney IP, Disney's going to like organize that and collect it and put it together in some sort of form for Disney Plus subscribers. So I don't know what that's going to look like. Anyways, the companies are framing the partnership as a testbed for, quote, responsible entertainment AI. No talent likenesses or voices included in the deal. There are also explicit promises around age appropriate use. Also safety controls for users. That includes a commitment to guardrails, blocking illegal or ([Time 0:03:08](https://share.snipd.com/snip/f7cf901d-62cc-4a1d-b994-aea0db428f2d)) - **Guardrails But Familiar Risks Persist** - Disney disallows talent likenesses and voices in the deal and emphasizes safety and age-appropriate use. - The hosts worry about recreating the "kid slop" era of unregulated children's content even if now legal and curated. Transcript: Jason Howell No talent likenesses or voices included in the deal. There are also explicit promises around age appropriate use. Also safety controls for users. That includes a commitment to guardrails, blocking illegal or harmful content. We're talking Disney here and generative AI. Disney is notorious for controlling its IP. Also, it's very child-friendly. And then you've got generative AI that is notorious for not following the rules and doing things surprisingly and unexpectedly. What are your thoughts on this deal? This is a big one. Jenn Cutter One, I did not think any kind of deal was happening anytime soon. Jason Howell That's why I didn't know what was going to happen eventually that these things were going to converge, right? Jenn Cutter I figured this was going to drag out to like middle of 2026. And here we are where it'll probably be out by then. Yeah. But also my first thought was, do you remember like the, when kid slop took over YouTube for a bit and like the Spider-Man Elsa, like whole kidnapping video, this was a big thing. And my brain said like, oh, did we just bring this back legally? Jason Howell You're right. I totally forgot about that. Like I filed that away in my, in the catacombs of my brain. ([Time 0:04:30](https://share.snipd.com/snip/22bb7bc6-76c9-4bbf-b696-5d65d4d2e78c)) - **Watermarks Can Imply Fake Partnerships** - Disney also sent Google a cease-and-desist accusing Gemini of training on Disney IP and misrepresenting partnership via a Gemini watermark. - Disney argues the watermark implies legitimacy and unfairly associates Google with Disney properties. Transcript: Jason Howell That's true. That's true. Now, along with this, Disney has sent cease and desist letter to google accusing it of copying a large set of disney films shows characters all that stuff to train its own gemini model You know nano banana and all that uh vo that sort of stuff disney says generated imagery using its ip has been illegitimately distributed on youtube and youtube shorts dis mentions The ability to generate figurines of its franchises using, you know, Nano Banana and Gemini, that sort of stuff. And I thought this was interesting. The fact that when you generate something with like Nano Banana, it automatically puts the little like Gemini star logo down in the corner to, as a hint, Google does this as a hint to say, This image was generated. And obviously it also, it also shares with Gemini. It's a watermark, right? But they framed it as this is a way of making sure that people don't fall for stuff that isn't real. And Disney is saying, no, when you do that, you make it seem like there's a legitimate partnership here between our IP and your product. ([Time 0:07:24](https://share.snipd.com/snip/e756949b-28a4-4a6c-b350-b5881ecb50ab)) - **Strengthen Chatbot Safety Controls** - The US state attorneys general warned AI firms about chatbots producing delusional responses that may violate consumer and mental health laws. - Companies should invest more in safety protocols and design systems that challenge harmful user delusions. Transcript: Jenn Cutter Microsoft, Meta, Google, and Apple are among the recipients of the letter. None have provided comments at this point. Jason Howell Yeah, so no teeth here, but it is like saying like, hey, we don't like it. There's, you know, there's a lot of people that have signed the letter basically just pushing back on the fact that, you know, AI is highly imperfect when it comes to, um, when it comes To the information that it spits out, a lot of people, kind of take it at face value. And then you go down the road of chatbots and chatbots are really good at kind of, I don't know, telling you what you want to hear. And so when you're talking about like – by the way, it was their words, the delusions thing. So just so you know, but if chatbots are feeding into a user's delusion, part of that is because these chatbots are just like, the way they are designed by and large is to not really push Back, not to counter. If I come in with an idea that I want to chat with a chatbot on, they really created these systems. So it's like, okay, yeah, we'll explore that with you instead of like, eh, buddy, maybe that's the wrong direction. Let's think about this. Jenn Cutter Yeah. And this is a government comment, basically, which is, you know, obviously it's stronger than all of the media coverage. So all of the media coverage got through to the attorneys general. And this is the official heads up. Jason Howell Yeah, I do think it's also kind of interesting that what they're saying here is, you know, consumer protection, mental health laws. And these products, I mean, there, you know, there was just another story today. I didn't include it in the rundown, but on TechMeme about, you know, a chatbot that someone had a long conversation with it or whatever. I didn't read the story entirely, but from the headline, you know, it's another one of those stories where it's like, you go down the rabbit hole with a chatbot, and then that leads you To the point of doing something unconscionable on the other side, you know, and it just it just goes to show not everybody is in a kind of like a self affiraffirmed, confident state where They can look at these tools through a critical lens and go, wait a minute, where am I headed right now? And, you know, part of that is on the human side of things. But I do feel that people, you know, those who are creating these tools, they carry responsibility in this too. It's, you know, it's, I think the responsibility lies on both sides as far as that. Jenn Cutter Well, to bring back another throwback, remember when GPS became ubiquitous and people would drive into lakes because they trusted the GPS more than their eyes. Jason Howell Totally. Like, yeah, you know, you still got to use common sense. You still got to use critical thinking. Yeah, that's a really good point. I'm happy you brought that up. I forgot about that. ([Time 0:13:41](https://share.snipd.com/snip/e0a9438a-4a0d-4ade-9258-51968960bccc)) - **Live Video To 911 For Faster Help** - Google launched live encrypted video sharing to 911 dispatchers on Android to enable real-time aid and potentially life-saving guidance. - The hosts praise integration of emergency features across Android devices as broadly beneficial. Transcript: Jason Howell Well, Google launched a new emergency live video feature for users running Android 8 or later. This is going to allow users to send live video from their smartphone to 911 dispatchers during emergencies. The new feature is going to appear when the user is on a call or a text thread with emergency services. And basically it allows dispatchers to send a request to share live video with a single tap on the other end. So it makes it very, very easy. Google says the new feature can be useful in scenarios where life-saving steps are needed before help arrives. So if you don't know how to perform CPR, maybe firing off the video gives some extra connection to that process. Could also, in my mind, make it a little more chaotic. But anyway, some information is better than no information. The live video is encrypted by default and the user can stop sharing at any time. I think this is a great feature. ([Time 0:16:31](https://share.snipd.com/snip/27f66784-0dfc-4d4c-9d41-dba34e84f952)) - **Rivian Builds Its Own Autonomy Chip** - Rivian announced an in-house 5nm self-driving chip made at TSMC and a lower-priced Autonomy Plus subscription for hands-free driving. - The moves mark Rivian's shift away from NVIDIA and aim to compete on price and availability across U.S. and Canada. Transcript: Jenn Cutter Vehicle maker Rivian Automotive made a couple of announcements at their first ever Autonomy and AI Day, including a new in-house self-driving 5-nanometer chip called the Rivian Autonomy Processor moving away from NVIDIA's processors. The new chips will be produced at TSMC and are expected to launch in late 2026. Rivian also announced a new driver assistance package called Autonomy Plus, which will be $49.99 a month as a subscription or $2,500 as a one-time fee. It includes universal hands-free feature for over 3.5 million miles of roads across the U.S. And Canada. This package is lower than Tesla's 99 a month subscription or $8,000 one-time fee for their full self-driving. ([Time 0:18:06](https://share.snipd.com/snip/50bca665-c1c4-4329-9b6f-b2f2416b71f6)) - **Gemini Gets Agentic Research And iOS Chrome** - Google upgraded Gemini with agentic deep research and integrated Gemini into Chrome on iPhone and iPad. - These moves turn Gemini into a proactive research tool and expand access across more devices. Transcript: Jason Howell Gemini, sorry for this to be a Gemini bus, but we're still on it. Gemini is now built into Chrome on iPhone and iPad. You know, it had come to Android, come to desktop a really long time ago. Now it's finally coming to iOS, bringing that spark icon that we were talking about earlier to the left of the address bar. I think it replaces the Google Lens icon that used to be there. But when you tap that new icon, you'll access the Pages tool and there you will find Ask Gemini and it integrates it to the browser and everything. ([Time 0:20:06](https://share.snipd.com/snip/3ecf9a52-cd3a-4c1a-b04b-f4eb00bf52cd)) - **Agents Will Market To Other Agents** - Fetch.ai positions itself as a directory and marketplace for AI agents so developers can mix and match prebuilt agents into workflows. - Andy Beach predicts agent-to-agent interactions will create a new "search agent optimization" economy. Transcript: Jason Howell Agents have come up already a couple of times on this show. AI agents are becoming the new shopping interface in a number of different ways. Andy Beach tells Tom Merritt why Fetch thinks it can organize the entire agent economy. Andy, thanks again for joining us. Hey, Tom. Good to see you. Tom Merritt So aside from all the jokes about making Fetch happen, explain what Fetch is. What is this company doing? Andy Beach Yeah, so they are thinking of themselves as sort of a new shopping interface and an agent economy system. They effectively are a portal with three products now launched that allow us to go build out the agentic workflows that we're talking about. So they are effectively a catalog of all the agents that have been registered with them for various tasks. And if I'm building out a workflow, I can go find the things I want rather than trying to build it myself to put together some sort of orchestration to do a variety of different elements. Tom Merritt If I've got this right, instead of me going into Anthropic and saying, hey, make me an agent to do this, I could go to Fetch and say, I want to do this and it would give me the agent. Andy Beach Yeah, exactly. Like maybe maybe you work in travel and you need a couple of different types of agents, one for pricing, one for inventory systems, one for policy control. And you don't want to go build one. You want to go look for somebody that has already built something and compare several different versions and then pull them into your workflow for a larger piece. Tom Merritt Yeah. So this is the search engine of agents, is it not? Andy Beach Yeah. In many ways, it is sort of that. And what's fascinating is, A, we just recently had the piece that we talked about around the LLM bubble. And I think this is an indicator and points to some of those trained agents and those trained pieces that Clem was talking about in that piece that will become the next wave and the pieces That are out there. What I think is interesting is this company is sort of taking the extra step of building out the catalog of them, but also what you really are going to have is an agent to agent conversation Where you're probably not searching it. You have an agent that is going and looking for the right things. And that agent needs to know how to talk to the other agents and interrogate it to find the ideal aspect. So we're almost moving to this B2B marketing that is agent to agent marketing. Tom Merritt So you don't want SEO, you want SAO, search agent optimization. Exactly. Andy Beach I think we're going to get a ([Time 0:22:32](https://share.snipd.com/snip/7f583c24-3c1b-4286-91f2-f602d6013b2d))