Skip to content
sobolewski.proAI today, every eveningSubscribe

Once a day. Unsubscribe with one click. Privacy policy

Claude Code won't build game clones anymore. Anthropic confirms it

Claude Code and game clones: Anthropic added two sentences against copies of known titles to the Opus 5.5 prompt, and developers report refusals.

Piotr Sobolewski4 min read

This article in other languages:العربيةDeutschEspañolFrançaisItaliano日本語Polski简体中文

In this article
  1. How does the game clone block in Claude Code work?
  2. Why is this getting so much attention?
  3. Does Claude also block reverse engineering?
  4. What’s new here in the context of AI?
  5. What should we watch out for?
  6. FAQ

Anthropic has confirmed that Claude Code on Opus 5.5 is meant to refuse building game clones, meaning recreations of specific, existing games. As Game Developer reports, a company spokesperson pointed the outlet to a post on X by Addy Osmani of Anthropic, which says two sentences were added to the Claude Code prompt to steer the model away from recreating other people’s games. On Hacker News this is already the fourth thread about the change since October 30.

How does the game clone block in Claude Code work?

The change was spotted by an X user who goes by Chasm. According to them, the new rule landed in Claude Code around October 27, under the internal name “tengu_heron_brook”, and it can’t be turned off. Game Developer quotes both parts of the rule:

  1. “Do build original games, games in a genre, and games from the user’s own designs and assets”: Claude should build original games, genre games, and games from the user’s own designs and assets.
  2. “Do not build a recreation of a specific existing game, even if it is renamed or called original”: it should not recreate a specific existing game, even if we rename it or call it original.

Osmani explains that original games, genre games and games built from our own designs “are meant to keep working”. The only reason given: using Claude to infringe on someone else’s intellectual property violates the usage policy.

Why is this getting so much attention?

Because “make me a clone of a famous game” is, for many of us, the natural way to test a coding agent, and now the model is supposed to refuse such requests, even if we rename the project.

The second reason is how the change was rolled out. The rule came from the server, with no changelog entry, and people learned about it from the model’s refusals. There is already issue #106167 on GitHub, in which the owner of Conflict Simulations LLC writes that Claude stopped helping with digital versions of tabletop wargames that they had been building for months for clients who own those games. The author asks for three things: narrowing the rule to actual infringement, an exception path for verified professional work, and documenting such changes in the changelog. At the time of writing, no one from Anthropic had replied there.

Does Claude also block reverse engineering?

This is where it’s worth slowing down. Some headlines (e.g. in Mashable and Tom’s Hardware) talk about blocking game “reverse engineering”, but the confirmed rule only covers recreating a specific existing game. Reports of other refusals, e.g. when analyzing a memory dump after a crash on Windows, come from users, mostly on Reddit, and as WindowsForum notes, nothing links them to the new rule so far. According to the same piece, in an XDA test Opus 5.5 agreed to help decompile Pokémon Diamond and drew the line between reverse engineering and recreating a game on its own.

So let’s take reports of a broader block with a grain of salt until Anthropic addresses them.

What’s new here in the context of AI?

The ban on copyright infringement itself isn’t new: Anthropic’s usage policy has long prohibited infringing on others’ intellectual property (the current version has an effective date of November 12, 2026). What’s new is that the company wrote a specific category, games, straight into the agent’s prompt, instead of leaving the judgment to the model.

This is another time we learn about hidden Claude Code behavior from users, just like with the hidden agent summaries that eat tokens. If you’re just getting started with Claude Code, I covered the basics in a tutorial.

What should we watch out for?

First, the boundary. A “game in a genre” is allowed, a “specific game” is not, and the model decides where the line falls between a retro-style platformer and a copy of a famous title. The first reports suggest it will get this wrong in both directions.

Second, project context. The author of the GitHub issue writes that the refusals persist across sessions on the same repositories, and one commenter suspects that session memory or Git history is to blame, both of which can be cleared.

Third, communication. So far we have a post by one person at Anthropic on X, which the company spokesperson points to, not an official announcement or a Release Notes entry.

FAQ

Will Claude Code still make a game?

Yes. According to Anthropic, original games, genre games and games from our own designs and assets should keep working. The block applies to recreating a specific existing game.

Does renaming the game get around the block?

It shouldn’t. The rule says outright that it also covers games that are renamed or called “original”.

Can the rule be turned off?

According to Chasm, who discovered it, no. The rule is delivered from the server to Claude Code on Opus 5.5.

Since when has the game clone block in Claude Code been in place?

According to Chasm, since around October 27, 2026. Game Developer added Anthropic’s confirmation to its story on October 30.

  • AI
  • Anthropic
  • Claude Code

AI today, every evening

The day's biggest AI news and new posts from the blog, every evening. Short, with sources.

Once a day. Unsubscribe with one click. Privacy policy