D&D General Hasbro CEO Says AI Integration Has Been "A Clear Success"

However "people make the decisions and people own the creative outcomes".
Copy of Copy of Copy of pODCAST358-fr (11).png


We've known for some time that Hasbro CEO--and former president of Wizards of the Coast--Chris Cocks is an avid AI supporter and enthusiast. He previously noted that of the 30-40 people he games with regularly, "there's not a single person who doesn't use AI somehow for either campaign development or character development or story ideas." In a 2025 interview he described himself as an "AI bull".

In Hasbro's latest earnings call, Cocks briefly addressed the use of AI within the company. While he mentions Hasbro, Wizards of the Coast and the digital studio teams, he doesn't specifically namecheck Dungeons & Dragons. However, he does tout Hasbro's AI integration as a "clear success", referring primarily to non-creative operations such as finances, supply chains, and general productivity enhancements, and emphasises that "people make the decisions and people own the creative outcomes". He also notes that individual teams choose whether or not to use AI.

So while it is clear that AI is deeply embedded in Hasbro's workflows, it is not clear to what extent that applies to Dungeons & Dragons. WotC has indicated multiple times that it will not use AI artwork, and its freelance contracts explicitly prohibit its use. The company also removed AI-generated artwork in 2023's Bigby's Presents: Glory of the Giants.

Before I close, I want to address AI, and how we're using it at Hasbro. We're taking a human-centric creator-led approach. AI is a tool that helps our teams move faster and focus on higher-value work, but people make the decisions and people own the creative outcomes. Teams also have choice in how they use it, including not to use it at all when it doesn't fit the work or the brand. We're beyond experimentation. We're deploying AI across financial planning, forecasting, order management, supply chain operations, training and everyday productivity. Under enterprise controls and clear guidelines around responsible use and IP protection. Anyone who knows me knows I'm an enthusiastic AI user and that mindset extends across the enterprise. We're partnering with best-in-class platforms, including Google Gemini, OpenAI and 11 labs to embed AI into workflows where it adds real value. The impact is tangible. Over the next year, we anticipate these workflows will free up more than 1 million hours of lower-value work, and we're reinvesting that capacity into innovation, creativity and serving fans. Our portfolio of IP and the creators and talent behind it are the foundation of this strategy. Great IP plus great storytelling is durable as technology evolves, and it positions us to benefit from disruption rather than being displaced by it.

In toys, AI-assisted design, paired with 3D printing has fundamentally improved our process. We've reduced time from concept to physical prototype by roughly 80%, enabling faster iteration and more experimentation with human judgment and human craft determining what ultimately gets selected and turned into a final product. We believe the winners in AI will be companies that combine deep IP, creative talent and disciplined deployment. That's exactly where Hasbro sits. As we enter 2026, we view playing to Win and more importantly, the execution behind it by our Hasbro, Wizards of the Coast and digital studio teams as a clear success.
- Chris Cocks, Hasbro CEO​

Wizards of the Coast's most recent statement on AI said "For 50 years, D&D has been built on the innovation, ingenuity, and hard work of talented people who sculpt a beautiful, creative game. That isn't changing. Our internal guidelines remain the same with regards to artificial intelligence tools: We require artists, writers, and creatives contributing to the D&D TTRPG to refrain from using AI generative tools to create final D&D products. We work with some of the most talented artists and creatives in the world, and we believe those people are what makes D&D great."

A small survey of about 500 users right here on EN World in April 2025 indicated that just over 60% of users would not buy D&D products made with AI.
 

log in or register to remove this ad

It is more about text than video, and I think the idea is that there is a continuum. The endpoints are:

1) "Give me an adventure set in the forgotten realms"

2) Writing your own mod

Somewhere in the middle is the prompting sequence:
-here is my original campaign setting
-here are the themes I'm trying to emphasize
-This is what the players have done so far
-they are interested in faction X. Suggest some ideas for adventures involving this faction.

Followed by human curation and development of those ideas.
Yes, this I'd what I'm talking about. In the setting I'm currently creating, I'm doing exactly this to see what ideas it pulls in that I haven't thought of to emphasize at certain intersections of factions and such and pitfalls and probabilities in my system and design before I ever get it to someone else to take a look at it. I'm using it in the exact same way I usually do with code to see how it turns out. The AI isn't writing anything, not really doing anything other than bouncing ideas around.
 

log in or register to remove this ad

I shouldn't do this, but ... this grognard is gonna have another AI discussion.
Disclaimer: I have been a content creator, as a professional writer/photographer. I mostly dislike AI, but I try to be reasonable. Let's get started.


So he uses an AI voice capture tool - like some of the campaign summary tools mentioned above - and captures his actual game play. He then uses AI editing tools to cut out the irrelevant chat and refine it into room descriptions, NPC goals and behaviors, item descriptions etc. before using these AI edited notes of his own sessions to write the actual adventure for publication.

This is nothing more than an automated transcript of what was said. It isn't "creating" anything. It is simply transcribing. This is supportable.

I wish WOTC would be consistent, just two years ago they went with the "we are not going to allow any AI art ...."

As a company their "moral compass" is all over the map.

They are a company. Their moral compass is "what will make our shareholders and CEO more money."

it’s not fuzzy, if the AI creates text that you the edit, you used AI. A spellchecker is not AI

A spellchecker is a form of AI.

When I give an art order to an artist and they create that illustration for me, I did not do the illustration. The artist did.

When you give an art order to AI and it outputs an image for you, you did not make that image, AI did.

Asking someone or something to do something is not the same as doing the thing. When I hire a plumber to fix my kitchen sink, the plumber fixed the sink, not me.
Bingo.

I have tried to use AI for D&D stuff. I even explored the tool to see if it could make my life any better. It's utter freaking garbage at anything other than "give me a name."

It's useful for names/words/areas/whatever, but not much more.

Oddly enough, the one issue I have with AI (that it creates total fabrications that look convincing) is that said hallucinations are useful, maybe, for gaming.

For example, I just cracked open an engine and wrote:

"Create a low-level NPC that the party might meet, one who has no connections to any established lore or plot, but is merely interesting."

I got three pages of mostly unusable crap. What it tried to create was the generic "filthy pre-pubescent male British street urchin who provides comic relief" stereotype. With an attack pigeon and a missing spoon stolen by a cat ("Would you like to craft an adventure for that?" I coulda cracked that off the top of my head in six tenths of a second. Except the stolen spoon. (I don't do those kinds of pharmeceuticals.)

EVERY time I try AI for creative purposes, it turns out utter *****. Like serious, utter unuseable crap. It's useless except for

"Give me 200 names, first and last, drawing from cultures that are equal to British, Welsh, Norse and Greek. Some Greek/Welsh Greek/British combinations are acceptable."

One of the big-name YouTube influencers admitted that he used AI tools. Stopped watching. Garbage, almost all of it.

Hell, Chat GPT can't even count to 200.
 


no it is not, never was, it is a stupid algorithm that compares what you typed against a list of words it ‘knows’ and maybe knows some basic grammar rules. They existed long before AI and do not work in the same way as AI at all.
Ok, So how would you argue that a modern AI spellcheckimifactor is different from the good old days?

I mean, totally off-topic now but this is the interwebs ....
 

For example, I just cracked open an engine and wrote:

"Create a low-level NPC that the party might meet, one who has no connections to any established lore or plot, but is merely interesting."

I got three pages of mostly unusable crap. What it tried to create was the generic "filthy pre-pubescent male British street urchin who provides comic relief" stereotype. With an attack pigeon and a missing spoon stolen by a cat ("Would you like to craft an adventure for that?" I coulda cracked that off the top of my head in six tenths of a second. Except the stolen spoon. (I don't do those kinds of pharmeceuticals.)

EVERY time I try AI for creative purposes, it turns out utter *****. Like serious, utter unuseable crap. It's useless except for

"Give me 200 names, first and last, drawing from cultures that are equal to British, Welsh, Norse and Greek. Some Greek/Welsh Greek/British combinations are acceptable."
I'm IT, and for my entire group we think it is mostly hype. We do have some professional uses of AI, and I have personally been involved in for around ten years now, but these are hyperfocused cases of pattern matching. Meanwhile, I've been working with it in cases some of our users want to use it. It being MS Copilot, because that's what our company gives us.

Unfortunately, there is something to "prompt engineering" that will get you better results. First, you have to tell it who what voice you are speaking as "Writing as a grognard dungeon master running 5E D&D..." and then you want to give it as many details as possible, such as describing the setting it is in, what details you want such as name, background, dark secrets, etc. In general, the more details you give it, the better your result will be. I'm prepping for a large sandbox campaign, so I've been trying to coax encounters and dungeons and their challenges out of it. Like I said, it will get better, but still won't be good for much more than brainstorming ideas. You can tell it the version you are running, and it will spit out some things that might make you think it actually understands, but it doesn't and will hallucinate game specifics. It has told me "Use Monster A (substitue Monster B)" which is a thing I have seen in 5E adventures. However the monster it will tell you to substitute will not exist in any Monster Manual or other online resource either. I've had it tell me to apply templates. It's even asked if I want a stat block for monster, and when I said yes, it gave me something unusable for even the most rules light OSR game. I also don't see how you got three pages out of it. I have to specifically tell it how many sentences I want to get anything large than one.

For more technical things, it certainly can't do coding of anything obscure or hard. I asked about a problem I was having and it ignored my request for code and gave me some cut and past text from two of the three websites I was able to find talking about the problem. So far, I have to have it kick back even a one line bash script that will actually run. It will make up named variables that don't exist and other such junk.
 

Ok, So how would you argue that a modern AI spellcheckimifactor is different from the good old days?

I mean, totally off-topic now but this is the interwebs ....
First of all I would argue that today's spellcheckers still do not use AI. Second, that the underlying mechanics of AI, or rather LLMs, are fundamentally different from what spellcheckers use. The latter or more looking up words in a list of allowed words and maybe some hard coded basic grammar rules while LLMs are basically calculating the probability of one word following another / taking the context it appears in into account.

Can there be some functional overlap with the AI side performing some spellchecking functionality, sure, but from a technical perspective no one would call that AI a spellchecker, they would still call it an AI.
 

Remove ads

Remove ads

Top