Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
Dan Houser, the former Rockstar Games co-founder and writer associated with Grand Theft Auto and Red Dead Redemption, has compared the possible spread of AI-generated training data to “mad cow disease.” Speaking on Virgin Radio UK’s The Chris Evans Show on November 26, 2025, Houser said AI could “eventually eat itself.”
He was using a vivid metaphor for a real technical concern: if future AI systems repeatedly learn from low-quality machine-generated material, errors, distortions, and sameness could accumulate. But Houser also acknowledged that his understanding of AI was “really superficial,” so his remarks should be read as cultural criticism rather than a formal technical forecast.
| # | Preview | Product | Price | |
|---|---|---|---|---|
| 1 |
|
DOCUMENTARY NOW: SEASONS 1 & 2 DVD | $8.21 | Buy on Amazon |
| 2 |
|
The Planet Earth Collection (DVD) | $19.89 | Buy on Amazon |
| 3 |
|
Michael Jackson's This Is It | $9.95 | Buy on Amazon |
| 4 |
|
Great Barrier Reef, The (DVD) | $7.30 | Buy on Amazon |
| 5 |
|
Titanic - The Definitive Documentary Collection + BONUS | $9.17 | Buy on Amazon |
What Dan Houser actually said
Houser appeared on the British radio program while promoting his science-fiction book, A Better Paradise: Volume One: An (Almost) Complete History. This Chris Evans is the British radio host, not the American actor.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteDiscussing the future of generative AI, Houser said:
#1 Best Overall
“AI is gonna eventually eat itself.”
He explained that AI systems draw on information found across the internet, while the internet may increasingly contain material produced by AI systems. He compared that feedback loop to “when we fed cows with cows and got mad cow disease.”
Houser added an important qualification: AI would perform “some tasks brilliantly,” but not every task brilliantly. He also described his own technical knowledge as “really superficial,” a caveat that matters when separating his opinion from established research.
Coverage of the appearance, including reports from GameSpot and PC Gamer, also highlighted Houser’s criticism of people promoting AI as a universal replacement for human creative judgment.
Recommended Free Tools
What “AI eating itself” means
Houser was not suggesting that artificial intelligence literally develops a disease. His analogy points toward concerns often discussed as synthetic-data contamination, recursive training, and model collapse.
Rank #2
- Generative models are trained on large collections of text, images, audio, code, and other data.
- More of the material published online is now generated or assisted by AI.
- If later training datasets include too much machine-generated content, new models may learn from earlier models’ approximations, mistakes, and stylistic habits instead of primarily learning from original human-created material.
- Across repeated generations, outputs could become less diverse, less accurate, or increasingly detached from the source material.
That is the technical concern Houser was gesturing toward. It is not the same as saying that every model trained with any synthetic data will fail. The outcome depends on the proportion of synthetic material, how well it is filtered and labeled, whether original data remains available, the task being modeled, and the quality of training and evaluation.
Synthetic data can also be useful. Carefully generated, filtered, and human-reviewed examples may help improve a narrow system. Some models are trained or fine-tuned on curated or proprietary datasets rather than indiscriminately scraping the open web. A system may become less reliable in one area without becoming completely unusable.
Why the mad cow comparison is both memorable and imperfect
Mad cow disease, or bovine spongiform encephalopathy, was associated with contaminated animal-derived feed practices that helped transmit abnormal proteins through cattle populations. Houser’s metaphor maps that idea onto an information loop:
- Cows consuming cattle-derived material: models learning from machine-generated material.
- Disease spreading: errors or distortions being repeated in later datasets and outputs.
- “Eating itself”: a system’s output becoming part of the input for future generations.
The comparison is rhetorically effective, but biologically imperfect. AI systems do not contract a disease, and synthetic-data degradation does not operate through the same mechanism as prion transmission. “Model collapse” is a conditional risk and an area of technical study—not a guaranteed industry-wide outcome.
Rank #3
- Own the entertainment event of a lifetime and get a rare, private look at the creative genius at work in Michael Jackson's: This Is It!
Houser is skeptical, not categorically anti-AI
It would be too broad to describe Houser’s position as a rejection of all artificial intelligence. He explicitly said AI could perform some tasks extremely well. His objection was aimed more at sweeping claims that AI can replace human creativity or provide a single solution to every problem.
He was particularly critical of executives and advocates who present AI as superior to human creative judgment. In remarks reported by PC Gamer, Houser questioned whether some of those advocates were sufficiently humane or creative. Those are Houser’s personal judgments and should not be generalized into a claim about every AI developer or company.
There is also a reported counterpoint: earlier coverage said that Absurd Ventures, the company Houser founded after leaving Rockstar, was “dabbling” with AI. The report also described Houser as warning that the technology was less useful than some companies claimed. That suggests skepticism toward overpromising rather than an absolute ban on experimentation. Because this account comes through secondary reporting, it does not establish a detailed company-wide AI policy.
Who is Dan Houser?
Houser co-founded Rockstar Games and became one of the major creative figures behind its narrative-driven games, including Grand Theft Auto and Red Dead Redemption. He left Rockstar in 2020 and is no longer a company executive or spokesperson.
Rank #4
- Factory sealed DVD
He later founded Absurd Ventures, the company behind the A Better Paradise transmedia project. His background explains why the comments attracted attention among game writers, developers, artists, and players—but it does not make him an AI specialist, nor do the remarks represent Rockstar Games or Take-Two Interactive.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Why the quote resonates in the games industry
Generative AI is being debated across nearly every part of game production:
- game writing and narrative design;
- voice acting, likeness, and performance rights;
- concept art and asset creation;
- localization;
- NPC dialogue and behavioral systems;
- quality assurance and testing; and
- development workflows and production costs.
For game companies, automation may reduce repetitive work or help teams explore ideas more quickly. For writers, actors, artists, and designers, the same technology raises questions about consent, attribution, compensation, originality, and whether creative decisions are being reduced to patterns optimized for scale.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Houser’s comments land particularly strongly because he is associated with large, story-heavy entertainment projects. Still, his credibility as a game creator is not evidence that his prediction is technically certain. It is better understood as a warning about what happens when commercial pressure treats creative work as interchangeable and when machine-generated material begins feeding back into the culture from which future systems learn.
Best Value
- Factory sealed DVD
Is this the “dead internet theory”?
Houser’s concern overlaps with fears about an increasingly synthetic web: a web containing more automated articles, images, posts, videos, and other material. It also overlaps with the so-called “dead internet theory.”
That connection should not be overstated. The available evidence does not establish that the internet is already mostly AI-generated, and the dead internet theory has not been proven. Houser’s point is narrower and more plausible: if synthetic material becomes common in future training sources, it may complicate the quality, diversity, and traceability of the data used to build new models.
The strongest reading of Houser’s warning
“AI will eventually eat itself” is a metaphorical prediction, not a scientific diagnosis. The defensible technical interpretation is that models trained on increasingly synthetic culture could amplify their own errors, biases, and repetitive patterns.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Whether that becomes a serious form of model collapse will depend on how future systems source, label, filter, and evaluate their training data. The presence of synthetic data alone does not guarantee failure. But the feedback loop Houser describes is a legitimate question for AI researchers and an especially important one for creative industries whose work may become both training material and the target of automation.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
