Days before the biggest game launch in years, the man in charge of Grand Theft Auto picked a fight with AI. In an interview with IGN after Take-Two Interactive’s latest earnings, CEO Strauss Zelnick called out Anthropic and OpenAI over AI tools that can quickly recreate existing video games. His line, that it “seems ironic on the one hand to stomp on someone else’s IP and hope that yours is protected,” spread across gaming and AI X within hours. The surprise: Zelnick says AI clones are not a financial threat. He calls them a creative one.
Quick Take
- What was said: Take-Two CEO Strauss Zelnick told IGN that all companies “should be protecting intellectual property, their own and everyone else’s,” and named Anthropic and OpenAI as companies “in the intellectual property business.”
- The context: AI coding tools have made fast decompilations, ports and clones of existing games easier. Grand Theft Auto VI launches on November 19, 2026 for PlayStation 5 and Xbox Series consoles, with PC later.
- Anthropic’s position: Anthropic added two sentences to Claude Code’s prompt on Opus 5.5 to steer it away from recreating “someone else’s specific game.” Original games, genre games and games from a user’s own designs are meant to keep working.
- OpenAI’s position: According to GamesRadar, OpenAI “seemingly has no such restrictions in place for now.” OpenAI has not commented publicly on Zelnick’s remarks.
- Why it matters: The fight over AI and copyright is moving from training data to output: what an AI tool should refuse to build.
In plain English
The boss of the company behind Grand Theft Auto says AI companies are being hypocritical. AI tools can now help people copy existing video games very quickly, and the companies that make those tools want their own inventions protected. He says this is not about losing money but about players getting the game the way its creators made it. One AI company, Anthropic, has already told its coding assistant to refuse to copy specific games. Another, OpenAI, has no similar public rule.
What happened
IGN asked Zelnick whether hackers using AI to clone games, and possibly bring GTA 6 to PC unofficially, are a threat. His answer, as published by IGN: “I don’t think it’s a financial matter, in terms of a threat to our business. I do think it’s a creative matter. It’s important to everyone, our developers and our consumers alike, to deliver the experience Rockstar envisions. And at the same time, of course, we protect our intellectual property vigilantly.”
He added: “We’re not in the least cavalier about them. Technology is advancing quickly. And at the same time, I do think that pretty much everyone wants the Rockstar experience as Rockstar intended.”
When IGN brought up Anthropic’s recent move to block game recreations in Claude, Zelnick widened the target. “I think all companies should be protecting intellectual property, their own and everyone else’s,” he said. “And if you’re in the intellectual property business, which Anthropic is and OpenAI is, then it seems ironic on the one hand to stomp on someone else’s IP and hope that yours is protected.”
On the Q2 fiscal 2027 earnings call, according to This Week in Videogames, Zelnick told investors that Take-Two is “utterly focused on enhancing our technical defenses and enhancing our legal activities appropriately,” while noting that “every entertainment business that exists is subject to some element of piracy.”
Background: Claude’s game rule
The comments land on a dispute that started in late October. Around then, an X user posted what they said was a new rule in Claude Code’s system prompt: “Do build original games, games in a genre, and games from the user’s own designs and assets. Don’t build a recreation of a specific existing game, even if it is named or called original.” Anthropic’s Addy Osmani did not confirm the exact wording, but replied that Anthropic had “added two sentences to Claude Code’s prompt on Opus 5.5 to steer it away from recreating someone else’s specific game because using Claude to infringe IP is against our usage policy.” The Verge reported the change on November 2.
The rule has already hit fan projects. GamesRadar reported on a developer of a Pokemon fan game who said “Claude models are starting to refuse work on Pokemon fan projects,” after spending “up to $2,000 expenses a month” on the project. GamesRadar also notes that some Steam developers cancelled demos or playtests out of fear their games would be cloned before release, and that retro games such as Mario Kart Wii and The Legend of Zelda: Link’s Awakening have seen AI-assisted recompilations.
Zelnick has previously said Rockstar did not use generative AI to develop GTA VI, while being cautiously supportive of the technology, This Week in Videogames notes. And the calendar adds some irony of its own: Anthropic’s planned Nasdaq listing is expected around the same week as GTA VI’s launch.
How AI game cloning works (and why it is suddenly easy)
Three overlapping things get lumped together as “AI cloning”:
- Decompilation and recompilation: turning a game’s compiled code back into readable source and rebuilding it for new hardware, the route to unofficial PC ports.
- Look-alike rebuilds: prompting a coding agent to build a game “like” a known title, with its maps, mechanics and characters.
- Fan games: new games built around existing characters, sprites and names, which Claude now increasingly refuses.
AI coding agents compress all three from months to days because they can read, translate and write large amounts of code without getting bored. That is exactly why the new frontier models (see Claude Fable 5.2) are attractive to cloners and worrying to publishers.
Pros and cons
For users
- Pro: Rules like Claude’s protect the games people buy from cheap copies flooding stores, and Anthropic says original and genre games still work.
- Con: Fan projects and preservation efforts get caught in the net, and a refusal can come mid-project after months of work.
For developers
- Pro: Indie developers worried about clones of their demos get at least one major AI tool that is told to say no.
- Con: The line between “a game in a genre” and “a recreation of a specific game” is fuzzy, and the rule is enforced by a model’s judgment, not a clear spec.
For companies
- Pro: Publishers gain a public argument: AI labs that guard their own IP should guard others’ too.
- Con: Prompt rules are easy to bypass with other models, including locally run ones. Take-Two itself is betting on technical and legal defenses, not on AI vendors’ goodwill.
What X is saying
The quote went viral through gaming accounts:
Take-Two CEO Strauss Zelnick called out OpenAI and Anthropic for allowing AI tools to recreate existing video games, arguing that tech companies should respect the intellectual property of game developers.
— Pirat_Nation (@Pirat_Nation) November 14, 2026 · View on X
The rule that triggered the question was Anthropic’s own:
We added two sentences to Claude Code’s prompt on Opus 5.5 to steer it away from recreating someone else’s specific game because using Claude to infringe IP is against our usage policy.
— Addy Osmani (@addyosmani) October 30, 2026 · View on X
Fan developers are already feeling it:
A Pokemon fan-game creator says they spent $2,000 a month and 10–14 hours a day on the project before Claude refused to help
— Dexerto (@Dexerto) November 14, 2026 · View on X
Developers say the new restrictions are hitting Pokemon fan games and ROM hacks by rejecting recognizable characters, sprites, names and logos
Replies split roughly three ways. Some turned the irony back on the AI labs: “Very ironic when you consider that the training sets used by AI companies is basically theft,” wrote @BoringBattlecry. Others blamed the cloners. Game developer @crierlon warned that people “cloning AAA games and bragging about it online is only going to ruin the AI coding agents for everyone else with safeguard hell.” And some pointed at the obvious loophole: “Locally running models remain undefeated,” posted @YourPalNyanus.
Grok’s take
In a reply summarizing the story, Grok wrote: “Take-Two CEO Strauss Zelnick recently told IGN that AI tools recreating existing games (like via Claude or OpenAI models) are a creative concern, not a financial threat to GTA 6. He called it ironic for IP-focused firms like Anthropic and OpenAI to allow cloning others’ work.”
Critical analysis
Zelnick’s argument is sharper than “AI bad.” He accepts that piracy exists and says it will be “manageable.” What he objects to is a double standard: AI companies defend their own models and code as valuable intellectual property, while their products make copying other people’s work trivial. That framing is hard for the labs to dismiss, because it does not depend on the unresolved legal fight over training data. It is about what the tools produce on request.
Anthropic’s answer is a prompt rule. That is fast and visible, but also brittle: it applies to Claude Code on Opus 5.5, it relies on a model judging what counts as a “specific existing game,” and it does nothing about other models. The Pokemon case shows the other side: a vague line catches fan creators who may be infringing but are not commercial cloners. OpenAI’s silence leaves it exposed to exactly the “ironic” charge Zelnick made.
Note also what Zelnick did not say. He did not announce lawsuits against AI companies or claim that Take-Two’s games were used to train these models. His remarks are a public warning shot, not a legal action.
What’s still open
- Whether OpenAI will adopt a similar rule for Codex or ChatGPT, or respond to Zelnick at all.
- Whether Anthropic will extend the rule to other Claude models and products, a question The Verge put to the company.
- How Take-Two’s “enhanced legal activities” will target unofficial GTA VI ports once the game is out.
- Whether courts will treat AI-generated look-alike games differently from human-made clones.
About the analysis
Researched and written with AI assistance under the editorial responsibility of the i10X News team. The title image is an AI-generated image. Read our AI Use Policy. Spotted an error? Request a correction.
This independent i10x analysis is based on IGN’s interview with Strauss Zelnick, earnings call quotes reported by This Week in Videogames, reporting by GamesRadar, Insider Gaming and The Verge, and public posts on X, including Addy Osmani’s statement for Anthropic.
- IGN: Take-Two boss Strauss Zelnick on GTA 6 expectations, GTA Online, and the decade to come
- IGN: Take-Two enhancing technical defenses and legal activities
- GamesRadar: Take-Two CEO calls out Anthropic and OpenAI over AI game clones
- Insider Gaming: GTA 6 AI cloning is not a financial threat, says Take-Two CEO
- This Week in Videogames: AI-generated GTA VI clones not a threat to our business
- The Verge: Anthropic clamps down on Claude’s AI video game clones
i10x Perspective
For two years the AI copyright debate has been about what models ate. Zelnick moves it to what models make, and that is a debate the AI labs cannot win with fair-use arguments alone. Anthropic has shown that a frontier lab can draw a line at “a specific existing game,” even if the line is blurry and fan creators get hurt. The pressure now shifts to OpenAI and everyone else shipping coding agents. Expect publishers to keep asking a simple question in public: if your tool will not help clone your own product, why will it help clone ours?
