Anthropic blocks Claude Code from cloning existing video games

The new restriction

Anthropic has confirmed a change to its Claude Code tool designed to stop the AI from rebuilding other people's video games. The update was discovered by an X user who noticed a new entry in the tool's ruleset under the name "tengu_heron_brook." This rule was pushed on October 27 and specifically targets the creation of clones.

The new instruction tells the tool to build original games, games within a specific genre, or games based on a user's own designs and assets. However, it explicitly forbids building a recreation of a specific existing game. This prohibition remains in place even if the user attempts to rename the game or call it something original.

According to reports from Game Developer, there is currently no way for users to switch this specific rule off. The change was implemented to ensure the tool does not facilitate the infringement of intellectual property. Anthropic did not issue a formal announcement but instead pointed to a post by a member of its technical staff.

Why the change happened

The timing of this update follows several weeks of reports regarding AI-assisted clones that unsettled independent game studios. Some studios even took drastic measures to protect their work from being fed into AI tools. For example, Digital Sun Games cancelled a public playtest for its game ReVamp to prevent someone from shipping a copy first.

Similarly, the studio Drunken Goddess Reflux opted for smaller private tests instead of public ones to avoid similar risks. These incidents have led to a broader conversation about how quickly AI can replicate existing titles. Developers have also pressured Valve to address how clones might displace original games on the Steam platform.

Anthropic's move is significant because it is the first action taken by an AI vendor itself to address this specific issue. While Valve acknowledged the problem during a recent event in South Korea, they stated they would not act as a gatekeeper for the tools developers choose to use. Anthropic's tweak provides a more concrete restriction on its own platform.

Technical implementation

The restriction was added as two sentences to the Claude Code prompt on the Opus 5.5 model. This model was recently launched by Anthropic with the goal of providing lower prices and stronger safeguards. The company has been tightening its usage policy multiple times over the past year to address various concerns.

The prompt serves as a set of instructions that guides how the AI behaves when a user provides a coding request. By including these specific sentences, Anthropic aims to steer the model away from recreating specific games. This is intended to align the tool with the company's usage policy, which prohibits using the service to infringe on intellectual property.

While the change is confirmed for Opus 5.5, it is currently unclear if every model within the Claude Code suite is affected. Users who find these restrictions too limiting have the option to switch to open-weight models. These models do not have the same built-in safeguards or rules imposed by a central company like Anthropic.

Limitations of the rule

Industry experts suggest that these safety limits may only deter the lowest-effort copycats. Researchers have frequently demonstrated that safety limits on generative AI can often be bypassed using rhetorical tricks. It remains difficult to determine how Claude will distinguish between a user's original idea and a protected game concept.

The line between a fan project, a homage, and a clone can be very blurry. Because of this ambiguity, Anthropic expects to continue iterating on the rules to find the right balance. The company has stated that it wants to hear about instances where a user is refused a request to build their own original game.

Despite these blocks, the rule is unlikely to stop all AI game clones on major storefronts. Other AI tools are not subject to Anthropic's specific rules, and Valve has maintained its stance on tool neutrality. Developers who face threats of cloning are currently being asked to report those incidents directly to Valve.

What comes next

Anthropic will likely continue to refine the balance between enabling creative work and preventing copyright infringement. As the technology evolves, the company may need to update its prompts to handle more complex requests. The goal is to allow developers to use the tool for original designs while blocking direct copies of existing titles.

The situation highlights a growing need for clearer standards in the AI and gaming industries. As more studios express concern over their IP, other AI vendors may follow Anthropic's lead. This could lead to a more standardized set of rules for how AI tools interact with existing creative works.

For now, the move serves as a specific safeguard for indie developers who fear their work will be instantly replicated. It marks a shift toward more proactive content management by AI providers. The industry will watch closely to see if these specific prompts are effective at stopping the trend of AI-generated clones.