Anthropic adds rules to stop Claude Code from cloning existing games

New restrictions for Claude Code

Anthropic recently introduced a specific regulation for its AI tool, Claude Code, to stop the software from generating replicas of existing video games. The update was reportedly launched on October 27 and was first identified by a user on X. This change is intended to ensure that the AI does not facilitate the creation of clones that mimic established titles.

The modification is internally referred to as "tengu_heron_brook." This specific instruction tells the model to prioritize the development of entirely new games or projects that fall within a specific genre. It also encourages the tool to work on projects that are based on the user's own original designs and assets rather than existing commercial products.

According to reports from WN Hub, this new regulation is non-optional for the people using the tool. Users cannot bypass this specific instruction to force the model to create a direct replica of a known game. This reflects a proactive stance by Anthropic to align the tool's capabilities with its broader usage guidelines regarding intellectual property.

Technical implementation and goals

Addy Osmani, a member of the technical staff at Anthropic, confirmed that this adjustment was added to the prompt for Opus 5.5. By embedding these instructions into the model's prompt, the company aims to guide the AI's behavior during the coding process. This ensures that the model's output remains within the boundaries of permitted creative activities.

The primary goal of this change is to prevent the tool from being used to infringe on the intellectual property of other creators. Anthropic wants to make sure that developers using the tool are not inadvertently or intentionally creating products that violate copyrights. The company stated that the aim is to support original and genre-specific creation using user-developed designs.

Osmani noted that the company will continue to refine this approach to find a proper equilibrium. This suggests that the current rules are a starting point in a larger effort to balance AI utility with legal safety. The company is looking for a way to provide helpful coding assistance without enabling the mass production of cloned content.

Industry concerns over AI cloning

This update comes at a time of growing concern within the gaming industry regarding the misuse of AI for cloning. Many developers are worried that AI tools could be used to quickly replicate the mechanics and aesthetics of successful games. These fears have led to several high-profile incidents involving major game studios and publishers.

For example, the studio Digital Sun recently delayed a public playtest for its game ReVamp. The company expressed worries that its demo could be inputted into an AI tool to be used for training or replication. This highlights the tension between sharing creative work and protecting it from being ingested by large language models.

Valve is also facing scrutiny over how it plans to handle AI-generated clones on its Steam platform. There are concerns that these clones could overshadow original games and harm the ecosystem for independent creators. Anthropic's new regulation is a direct response to these types of industry-wide anxieties about the ease of replication.

Challenges in enforcement

While Anthropic has implemented these rules, some industry observers question how practical they will be in the long run. Critics point out that restrictions on AI can often be bypassed by users who know how to craft specific prompts. If a user is clever enough to describe a game without naming it directly, the model might still produce a very similar result.

Furthermore, the tool currently lacks a mechanism to verify the ownership of the designs or assets that a user provides. If a user uploads a design that they do not actually own, the AI may still use it to build a game. This creates a technical gap between the company's intentions and the actual capabilities of the software.

The company is currently working to find a balance that allows for creative freedom while maintaining safety. Because the AI cannot