Nintendo Denies Lobbying Government on Generative AI | 08/10/25
Plus, vote on a newsletter topic coming this December!
Amazon’s changes to Luna include an AI-native Snoop Dogg(???)
Nintendo denies lobbying the Japanese government, but what’s the real story?
OpenAI make work for themselves, and announce another cashless collaboration.
Kingmakers is delayed to avoid relying on ‘fake frames’
Exciting newsletter topics and case studies coming your way!
First reveal of new AI and Games merch to premium subscribers!
Our monthly sponsor issue of the newsletter provides what you come to expect every week from AI and Games. Plus a deep-dive into what’s to come that is exclusive to our premium subscribers. Ranging from future newsletter topics, to YouTube episodes, new projects, our conference planning and much more.
Greetings one and all and welcome to this week’s edition of the AI and Games newsletter. Once again I am on the road, this time at Pocket Gamer Connect in Helsinki. So yes, greetings from Finland, and if you’re reading this in the UK or any farther west then I will have already delivered my session at the event, and am now preparing for a client visit tomorrow.
We have a bunch of news stories this week, and for premium subscribers is the big update on everything we have coming up in the next few months. Plus we hit a subscriber goal, so you can now vote in a newsletter topic you’d like to see before the end of the year.
All that and, over the paywall we announce for the very first time a really fun little secret that we’ve been cooking up for the past six months, Something you can get your hands on yourselves later this year.
Alrighty, let’s do this!
Follow AI and Games on: BlueSky | YouTube | LinkedIn | TikTok
Announcements
Alrighty, we’ve got some AI and Games Conference announcements, and we hit our subscriber goal - and so we want to hear from you on what you want to see in December?
AI and Games Conference 2025
It’s now less than a month until we’re back at Goldsmith’s University in London. Are you nervous? I’m nervous! I’m also excited because of our pretty exciting line-up. So let’s highlight three more talks coming your way.
But Why Is It Saying That?? Making LLM Powered AI Bots a Bit More Reliable
Batu Atyemiz / Fundamental Labs
As has been discussed in various places, including this great publication, generative models are great and all but they’re rather unreliable in practical situations. Batu will be highlighting his experiences of shipping an LLM powered agent within a Roblox game that has thousands of active users a day, and what you can take from this to apply to your own projects.
Debugging Across Time and Platforms: The Power of Determinism
Richard Kogelnig / Havok
We welcome our friends for Havok who have joined us this year as a partner sponsor of the event. Richard will be digging into an interesting challenge game AI developers face when shipping games on different hardware platforms: ensuring consistency of AI behaviour. Nothing worse when you have a bug that only exists on one console, and not the other. So we’ll be hearing more about Havok’s efforts in creating cross-platform determinism to make it easier to ensure if one platform has a problem, then all of them do.
Good Enough AI: Pragmatic Approaches for Creating Compelling Player Experiences in Games
Gaetan Soppe Drenjanin / Sports Interactive
Last but not least for this week, we welcome Gaetan from Sports Interactive, who is going to give an overview on how to build game AI in smart and practical ways. Leaning on his experiences on shipping NPCs and other gameplay systems in large AAA games, he will share that sometimes ‘good enough’, is all you need.
Vote for a Newsletter Issue in December
Back in August we announced our broader goals for this here publication and launched a subscriber drive to encourage more readers both to check us out, as well as support as a premium subscriber. You can find out more about that here.
A week or so back we quietly snuck past our first subscriber goal, with over 3500 people now signed up to receive this newsletter in their inbox. Thank you all so much for your support!
With this goal, we now open up the option to have one newsletter story a quarter be picked by you. So we’ve gone to ground to work on three stories that we think you’d like to hear about, and can vote on right here. Let me break them down for you…
AI-Native Game Performance on Steam
As discussed in previous issues, AI-native is the term given to games that use some form of generative AI as a core part of the gameplay experience. Now I don’t like that term - it’s designed to ignore the 40 years of existing work in AI for games - but that’s what the tech-bros are calling it.
But the big question is, while AI-native games are slowly appearing on Steam, are people actually playing them? So let’s dig into the data that’s out there and paint a picture of what size of market these games have?
Deconstructing AI Adoption Surveys
It’s become a bit of a twisted hobby of mine to look at surveys on AI adoption in the games industry and pull them apart to highlight their design flaws and editorial intent. There’s that quote attributed to Mark Twain that sticks in my mind: “facts are stubborn things, but statistics are more pliable.”
I fancy taking a look at all of the surveys on AI deployment in games in the past year or two and figuring what, if anything, can be gleamed from all this data.
What the EU AI Act Means for Games
I’ve spoken a few times about the broader implications of the EU AI Act, and what it could mean for games. But I fancy maybe digging a little deeper. How does this impact games studios? Why is this relevant to everyone regardless of what country you’re based in? What issues should a studio be addressing today to be ready?
All of this coming from someone who isn’t a lawyer of course!
Vote Now for What You Want to See
And so now it’s all in your hands: decide which of these stories you’d like to read! Survey will remain open for one week.
AI (and Games) in the News
Last week we had an issue dedicated solely to the news, so I’m glad things are a little bit quieter this week. A little bit anyway…
xAI is Hiring Video Game Tutors
xAI, creators of the antisemitic, holocaust-denying, misinformation-spreading, hate-speech-filled large language model ‘MechaHitler’, aka Grok - and it’s human avatar Elon Musk - are on a hiring spree for ‘AI tutors’. There’s one in particular looking for people who can help it understand video games.
The listing which at the time of writing is still open on LinkedIn, is looking for individuals that “will contribute to xAI’s mission by training and refining Grok to excel in video game concepts, mechanics, and generation”.
If you wonder why they are hiring for tutors, and why it doesn’t pay all that great compared to actual engineering jobs - paying somewhere between $40 and $100 an hour - it’s because this is a vital component of building large language models (LLMs) that is reliant on grunt work. LLMs need refined once they’ve been trained to get better at specific tasks, or iron out issues with them. This is an emerging space in AI development that is essentially another part of the broader service industry. A sort of the modern era equivalent of call centre or data entry work. You sit with an LLM and engage in conversation with it to refine it to be less error prone, and those interactions (unlike those of a typical user) are going to help update the behaviour of the system.
While Business Insider really bigs it up in this piece to sound much more glamorous than it is, it’s increasingly a loveless job to help steer these systems to they make mistakes less than usual.
So yeah, if you need the money and are happy to work for the UK’s most hated brand, then check it out!
Snoop Dogg has an AI Video Game Now?
Amazon announced last week they’re planning a big reset and rebuild for ‘Luna’, which is their cloud-based game streaming platform. Over the past couple of years it has seen a consistent stream of big titles come to the platform, but now it’s pivoting to focus slightly less on this, and more on party games?
Once you read the announcement, you can see the logic behind what they’re doing. What was once introduced as a competitor to Xbox, PlayStation and Nvidia’s efforts is now deviating away from the crowded ‘gamer’ market, and instead leaning more into casual and party titles that are more accessible to families. Essentially the big screen equivalent of that which permeates mobile spaces.
While they’re providing access to popular titles like Angry Birds and Ticket to Ride, they’re introducing a new title, GameNight, developed by Amazon themselves. In amongst it all is an AI-native game called Courtroom Chaos starring Snoop Dogg, in which players must improve their way around what is no doubt a serious but fun adjudicator.
Y’know the weirdest part of this story, is I’ve already seen AI-native games with Snoop Dogg in them behind closed doors this year. Clearly he’s all in.
Game That Could Not Be Conceived Without AI Is Made Without AI
So yeah, a story I ignored a few months back was a post on Twitter (you see why I ignored it huh?) in which user @de5imulate had showcased a concept for a pixel art style renderer in a game using Midjourney. With many arguing this was a real breakthrough and highlighted the future of AI in game development. You can see the video below if you wish to refrain from logging onto the hellsite.
Now the thing is I ignored it because I just don’t have the capacity to put up with such nonsense. You can’t built an entire game in this manner using generative AI - and even the original author said as such in an interview. Desimulate then decided to launch a games studio (OMW Game Studio), with the intent to try and make it real.
But yeah… that’s something people have already explored in game development to various degrees. This original post led to some pretty salty responses in gaming press. But more importantly, someone was already working on this style and beaten him to it.
You can wishlist Project Shadowglass now on Steam.
Nintendo Uncharacteristically Responds to Accusation of Lobbying Japanese Government Over Generative AI
A weird story that popped up over the weekend, was that a member of Japan’s House of Representatives, Mr Satoshi Asano, had stated on Twitter/X/Hellsite that Nintendo was “engaging in lobbying activities with the government” in relation to generative AI legislation.
Japan is currently going through an assessment of how it fits AI within broader society and governance. In fact the recent publication of the “AI Promotion Act” this summer details their government’s intents to go all-in on AI adoption. In this act they actively encourage the adoption of (generative) AI throughout society as a means to improve efficiency. Perhaps fitting with the culture, it does not have strict rules on the exploitation of intellectual property with generative AI - something that is generally permissible. Rather they have adopted a ‘name and shame’ policy, whereby it is expected individuals or organisations who operate in ways deemed unacceptable by the broader populace and the government are called out publicly - with potential action by politicians afterwards. This might sound outright bizarre in western circles, but reputation, notably corporate reputation, is a huge aspect to the success of businesses in Japan, and there’s a risk of losing trade and customer interest by deviating from the cultural norms.
Now we know that Nintendo have little interest in generative AI, with their stance largely unchanged since we reported on it last year. So this might give us context as to why Nintendo might lobby the government: they perhaps want stricter rulings on IP protections.
But perhaps the most interesting aspect is that Nintendo have - rather uncharacteristically for them - actually responded to this story on the hellsite, with a formal denial being published. Quoted below:
Contrary to recent discussions on the internet, Nintendo has not had any contact with the Japanese government about generative AI. Whether generative AI is involved or not, we will continue to take necessary actions against infringement of our intellectual property rights.
This has since led to Asano apologising and retracting his original statement. He then posted on Twitter saying:
Regarding the content I posted the other day, there were parts that contained incorrect information, and I have made the following corrections. I regret and reflect upon my insufficient verification of the facts. I sincerely apologize once again to all parties concerned and will endeavor to prevent this from happening again.
Furthermore, since the original, incorrect post was observed to be spreading again after it trended on X (formerly Twitter), I have just deleted it to prevent the further dissemination of false information and the resulting inconvenience. I will take this experience as a valuable lesson and apply it going forward.
Satoshi Asano
I think this is pretty interesting in that Nintendo perhaps don’t want to be seen to engaging on this topic behind closed doors, or at least don’t want it to be known that’s what they’re doing given the aforementioned corporate culture and the idea that they’re stance is kind of opposing a lot of what is expressed in the government’s promotion act. But hey, it’s all nicely timed given all the OpenAI guff…
OpenAI’s Had an Interesting Week…
Two stories of note here…
OpenAI Releases Sora2 using Opt-Out Model, Changes Mind
So yeah…. OpenAI announced their new Sora2 video model last week. During which they stated they’re going the route of making any and all copyright violations permissible, whereby companies would have to opt-out in order to ensure their IP is protected. Surprising absolutely nobody, the internet embraced this whole heartedly and started posting all sorts of videos that clearly violate a range of intellectual properties. I won’t be sharing them, as I don’t like giving it exposure, but I’m sure you’ll find them pretty quickly with a Google search or two.
Now in a world of dumb ideas this felt exceptionally brazen. Given as we’ve discussed in previous issues one of the reasons big corporations are going after AI company Midjourney in the courts is because they decided not to put any guardrails on their interfaces. This means that users can easily create IP violating materials, and it’s thanks to OpenAI’s guardrails on GPT and DallE that you can’t do this in their models.
What was OpenAI thinking? That companies would just magically get on board? Well what a difference a couple of days makes. With OpenAI’s Sam Altman writing a blog post last week stating they’re going the other route and now stating IP holders need to opt in. Giving them “more granular control” of the use of their IP.
I wonder how many outside organisations, and which ones, came knocking on their door demanding the changes.
New Partnership Between AMD and OpenAI Announced for Data Centres
GPU manufacturers AMD have announced a new partnership with OpenAI to support their deployment of new datacentres to power models such as GPT, DallE and Sora.
This, like many of OpenAI’s deals of late, is built without any clear exchange of actual cash. OpenAI will apparently purchase 6 gigawatts worth of GPUs, which should amount to tens of billions of revenue for AMD, but how and where they do that isn’t made clear. Rather it will just manifest somehow. Meanwhile AMD will provide stock to OpenAI that will vest based on specific milestones and targets being hit - amounting to around 10% of the value of the company.
AMD gains from being a new GPU partner for OpenAI - at a time where Nvidia makes most of its revenue selling GPUs to big AI companies. Plus OpenAI gets to make a big announcement that they’re shelling out cash to support the AI economy, without actually committing a specific cash amount.
Kingmakers Delay Announcement Makes a Dig at DLSS Adoption
Redemption Road games, creators of the upcoming Kingmakers have announced that the game, due out this week, has been delayed indefinetely until they address issues with the product. The social media post can be found below.
Now I found out about this because my friends and I are quite excited for this game. It looks dumb fun! But as I read through it, while they talked about the NPC AI, there was a different part of this that really caught my eye:
With Kingmakers, we have set out to push the Unreal Engine 4 codebase to its absolute limits, while still providing true 60fps to midrange PCs, without the need for fake frames. We are an 80% engineering team, who got into the business to push technical barriers.
The point about ‘fake frames’ is feeding into the discourse on the necessity to adopt AI upscaling such as AMD’s FSR and Nvidia’s DLSS as a means to ensure base levels of performance. A point I dug into in our recent issue of The Take.
Not surprising that studios are trying to avoid falling into the same trap as the likes of Borderlands 4 at launch and working to do the engineering work to get that stable performance out the gate. Best of luck to them.
The Sponsor Update
Alrighty, next up we’re going to dig deep into the work we’re doing behind the scenes. With an update on all of our various aspects of the business, what content is coming your way, and some conference surprises too!
As always, below the line is accessible to our wonderful premium subscribers.
Keep reading with a 7-day free trial
Subscribe to AI and Games to keep reading this post and get 7 days of free access to the full post archives.