Anthropic Steers Claude Code Away From Recreating Existing Games
Anthropic has added instructions to Claude Code that discourage users from asking the coding tool to rebuild someone else's game. The change surfaced after a user spotted a new rule in Claude Code's configuration, which was reportedly pushed on October 27. It lands while independent developers worry that AI tools make it easy to copy their games and demos. Anthropic describes the change as a way to steer the model's behavior rather than a guarantee that clones are impossible.
What the rule says
The rule appears under the label "tengu_heron_brook." It tells Claude Code to build original games, games in a genre, and games based on the user's own designs and assets. It tells the tool not to build a recreation of a specific existing game, even if the project is renamed or described as original. The X user who found it, known as Chasm, reported that there is no setting to turn it off.
An Anthropic spokesperson pointed Game Developer to a statement from Addy Osmani, a member of Anthropic's technical staff. Osmani said two sentences were added to Claude Code's prompt on Opus 5.5 because using Claude to infringe intellectual property is against Anthropic's usage policy. He added that original games, genre games, and games built from a user's own designs and assets are meant to keep working. He also said Anthropic will keep adjusting the balance and invited users who are wrongly refused to contact him. We found no separate Anthropic announcement or newsroom post.
Why it matters now
The most recent Steam Next Fest reportedly showed roughly double the number of demos of the previous event, with AI-assisted games accounting for most of the increase. Several studios, including Digital Sun Games, the developer behind Moonlighter, have canceled public playtests for fear that someone could feed a demo into an AI tool and release a competing version first. At a Valve event in South Korea, one developer also described emails threatening to copy a game with AI and release it first unless the developer paid.
Valve addressed the issue at Steam Indie Connect in Busan on October 27. Representatives acknowledged that generative AI has made an older cloning problem easier to scale, but said Valve does not plan to ban specific tools. Valve's Tom Giardino said making Steam "a smarter and more dynamic platform" is the starting point. No new AI policy was announced, though Valve said it wants to hear about developers being threatened. Valve also advised developers to build communities and keep updating their games.
What it can and cannot do
Anthropic controls Claude Code's behavior. It does not control Steam, and it does not control other AI models. Game Developer noted that the rule may deter only the most low-effort copycats, since safety limits on AI tools have repeatedly been bypassed with rhetorical tricks. It is also unclear how the software could judge whether a user owns a game idea or its assets. Commenters pointed out that determined cloners could turn to open-weight models with no such restrictions. That last point is the reporting's concern rather than an Anthropic claim, but it shows the limits of a single vendor's guardrail.
Open questions
The reporting leaves several practical questions unanswered. It does not explain where the line falls between a "game in a genre" and a recreation of a specific game. It does not say how the rule treats legitimate work such as preservation, modding, or interoperability. It does not say whether Anthropic intends the rule to be permanent. For now, developers worried about clones should not treat the rule as protection, and users who are wrongly refused have been told to report it to Anthropic.
Editorial note: this article was researched and drafted with help from an AI assistant made by Anthropic, the company at the center of this story, and then edited for publication.
Sources
- Game Developer (independent; relays Anthropic's confirmed statement)
- PC Gamer (independent)
- Inven Global (independent; report on Valve's Steam Indie Connect remarks)