Claude Code Draws A Line On Recreating Existing Games
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: Claude Code Draws A Line On Recreating Existing Games on ThorstenMeyerAI.com

Age 18–24?Offer from Amazon

Prime made for students and young adults

  • Fast, free delivery for dorm and study essentials
  • Prime Video and Amazon Music included
  • Member-only deals
Try Prime for Young Adults Free trial for eligible 18–24 year olds
As an affiliate, we earn on qualifying purchases.

TL;DR

Anthropic has added an instruction to Claude Code telling it not to recreate a specific existing game, even under a different name. The company says the instruction applies its existing intellectual-property policy to game development; original games and genre-based projects remain allowed. How the rule handles licences, permission and client work is not yet clear.

Anthropic has added a game-specific instruction to Claude Code, telling its AI coding agent not to recreate a particular existing game, even if the project is renamed or described as original. The company said the instruction applies its existing policy against intellectual-property infringement to game-building requests, while still permitting original games, genre-based projects and games built from a user’s own designs and assets.

The instruction reads:

“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 renamed or called original.”
XDA reported finding the text in Claude Code version 2.1.314. The source report says it also found the same wording in versions 2.1.314 and 2.1.316, but not in 2.1.312. Anthropic confirmed the change on October 30.

An Anthropic spokesperson referred Game Developer to a statement from Addy Osmani, a member of the company’s technical staff. Osmani said the instruction steers Claude Code away from “recreating someone else’s specific game” because using Claude to infringe intellectual property violates Anthropic’s usage policy. That policy already barred intellectual-property infringement; the reported change is a specific instruction for game requests inside the coding tool.

The immediate boundary described by the instruction is between a request to make a particular existing game and one to make an original game or a game in a broad genre. The source report says that, in XDA’s tests, Opus 5.5 still offered help with decompiling a game from the user’s own copy. That example does not establish how Claude Code will respond to every request involving reverse engineering, licensed material or permission from a rights holder.

At a glance
updateWhen: Confirmed October 30, 2026; reported in…
The developmentAnthropic confirmed a new game-specific instruction in Claude Code that directs the coding agent away from recreating existing games.
Thorsten Meyer AI
The rule · UTC

Infographic 01 · Claude Code, 27–31 October 2026

Two sentences, no off switch.

The exact wording in the Claude Code program, which versions carry it, and how it became public. Times in UTC.

2sentences added to Claude Code’s prompt, per Anthropic
2.1.314first Claude Code version with the text in the program (30 Oct)
0mentions in the public Claude Code changelog (checked 31 Oct)
1reason Anthropic gives: its usage policy bans IP infringement
Do“Do build original games, games in a genre, and games from the user’s own designs and assets.”
Don’t“Don’t build a recreation of a specific existing game, even if it is renamed or called original.”
  • 2.1.31227 Oct · text not in the program
  • 2.1.31430 Oct, 00:12 UTC · text in the program
  • 2.1.31631 Oct · text still in the program
  • ~ 27 OctRule delivered from Anthropic’s serversReportedFlag “tengu_heron_brook”; “no setting to turn it off”, per Chasm
  • 30 Oct · 00:12Claude Code 2.1.314 publishedVerifiedXDA and our own check find the two sentences in the program
  • 30 Oct · 07:01The rule goes public on XXChasm posts the full text
  • 30 Oct · 09:25First complaint on GitHubGitHubA board-wargame studio reports refused client work
  • 30 Oct · eveningAnthropic confirmsAnthropicVia a spokesperson to Game Developer, pointing to Addy Osmani’s statement

Sources: own check of the Claude Code 2.1.312, 2.1.314 and 2.1.316 program files (31 Oct 2026); npm version dates; Chasm on X, 30 Oct; XDA, 30 Oct 2026; GitHub issue #106167; Game Developer, 30 Oct 2026 (with Anthropic’s confirmation).
Powered by Thorsten Meyer AI

A New Limit on AI Game Prototypes

For developers, the change makes a familiar intellectual-property risk a direct product constraint: a request to reproduce a named game may now be refused by Claude Code, not just raise questions about how the resulting work could be used. That may affect rapid prototypes, browser demos and work commissioned by clients, particularly when a project closely follows an existing title.

The rule also arrives amid concern about public game demos that resemble established releases. The source report says studios Digital Sun and publisher Alliance Arts pulled or restricted public demos in October, with Digital Sun citing the risk of making a large build public months before launch. Those decisions were reported separately; the available material does not show that Anthropic’s instruction caused them.

For players and rights holders, the policy signals that at least one major AI coding tool is attempting to limit requests to recreate specific games. It does not settle whether a particular game is legally infringing, nor does it prevent people from using other tools. The practical effect will depend on how consistently Claude Code interprets the distinction between a broad genre and a close recreation.

Bose QuietComfort Wireless Noise Cancelling Headphones, Sandstone

Bose QuietComfort Wireless Noise Cancelling Headphones, Sandstone

Vitamix Propel Series 510 Professional-Grade Blender, 3 Programs, Black

Vitamix Propel Series 510 Professional-Grade Blender, 3 Programs, Black

As an affiliate, we earn on qualifying purchases.

From Fallout-Like Demo to Policy

The new wording followed a month of attention to AI-assisted game recreations. At the end of September, a creator posted a browser game set in New York that resembled Fallout. The creator described quests, dialogue trees, a V.A.T.S.-like feature and a working Pip-Boy, and said it was built with Claude Opus 5.5. Kotaku reported that the post drew more than 3.5 million views. Three days later, the creator said they had received a cease-and-desist letter from Bethesda and abandoned the project. The source account says the letter was shown as a screenshot; the material provided does not independently establish its contents.

Other AI-made ports and demos also circulated during October. Kotaku reported browser versions of Halo, Grand Theft Auto: Vice City and The Simpsons: Hit & Run. Reports of these projects do not establish that each was made with Claude or that each prompted Anthropic’s policy change. The company has not identified a specific game, publisher complaint or legal dispute as the reason for the instruction.

Anthropic confirmed the instruction after it was spotted in Claude Code. The source report says it was delivered from Anthropic’s servers around October 27 and appeared in version 2.1.314, released October 30. It also notes that OpenAI’s and Google’s usage policies prohibit intellectual-property infringement generally, but that the report found no comparable game-specific instruction from either company as of October 31, 2026.

“The instruction steers Claude Code away from “recreating someone else’s specific game” because using Claude to infringe intellectual property is against the company’s usage policy.”

— Anthropic technical staff member Addy Osmani, in a statement referred to Game Developer by an Anthropic spokesperson

How Permission Fits the Rule

Anthropic has not explained how the instruction treats licences or explicit permission from a game’s rights holder. The text refers to recreating a specific existing game but does not set out an exception for authorized work, adaptations or projects made for a client that owns the relevant rights.

A developer who makes digital versions of board wargames for clients who own those games reported that Claude Code was refusing paid work, according to the source report. The developer filed a GitHub issue, and Anthropic had not answered it at the time described. That report raises a practical question about the rule’s scope; it does not establish how the tool will handle all licensed projects.

The legal boundary is also fact-specific. The source report says copyright generally protects a game’s expressive elements, such as its appearance and sound, rather than its broad ideas or rules. It cites a 2012 US court ruling that found a Tetris clone infringing despite the absence of copied code. That example is relevant context, not a ruling on the Claude Code instruction or any of the projects mentioned here. Anthropic has not said whether it will publish further criteria or explain how users can challenge refusals.

Watch for Policy Clarification

The immediate next test is how Claude Code applies the wording to borderline prompts: requests for genre conventions, close visual or mechanical similarities, licensed adaptations and work using a user’s own assets. Anthropic’s public confirmation establishes the instruction’s purpose, but the source material reports no detailed guidance, appeals process or schedule for further policy updates.

Developers affected by refusals may seek clarification through Anthropic’s support channels or the GitHub issue cited in the report. Until the company provides more detail, teams planning game prototypes can distinguish their own designs and assets from requests that name or reproduce a particular title, while checking rights and permissions independently. The instruction is a constraint on Claude Code’s responses, not a legal determination about a project.

Key Questions

What did Anthropic change in Claude Code?

Anthropic confirmed an instruction directing Claude Code not to recreate a specific existing game, even if the user gives it a new name or calls it original. It says the tool can still help build original games, games in a genre, and projects based on users’ own designs and assets.

When did the instruction appear?

The change was confirmed on October 30, 2026. The source report says it found the wording in Claude Code versions 2.1.314 and 2.1.316, but not in version 2.1.312.

Does the rule prohibit making any game inspired by another game?

The text permits games in a genre but bars recreating a specific existing game. Anthropic has not provided a detailed test for where broad inspiration ends and a recreation begins, so the treatment of close or borderline examples remains unclear.

Can Claude Code help with a game if the user has permission?

The instruction does not specify how licences or rights-holder permission affect its response. A developer reported refusals involving paid work for clients who owned the games, and the source report said Anthropic had not yet answered the related GitHub issue.

Did the Fallout-like project prompt the rule?

Anthropic has not said what prompted the change. The source report describes the project and a creator’s statement that a Bethesda cease-and-desist letter led them to abandon it, but it does not establish that Bethesda’s action caused Anthropic to add the instruction.

Source: ThorstenMeyerAI.com

You May Also Like

The policy menu. There’s no single answer. There’s a menu — and choosing is a values choice in disguise.

Exploring the diverse policy options for addressing AI-driven economic changes, emphasizing values over technical solutions amid uncertainty.

AI’s First Cyberattack Was Never Supposed To Happen—It Was A Mistake

OpenAI’s AI models unintentionally launched the first documented fully autonomous cyberattack, revealing new security risks in AI development.

Anthropic Models Meet In-Region Inference On Amazon Bedrock

Anthropic says its models on Amazon Bedrock can run in-region in Seoul and Singapore, but model eligibility and data-handling details remain unspecified.

World Model Readiness: Are You Ready for AI That Acts?

Assess your organization’s readiness for AI systems capable of prediction and action with the new diagnostic tool, highlighting current capabilities and gaps.