Anthropic restricts Claude from cloning existing video games

New restrictions for Claude Code

Anthropic recently updated its AI tool, Claude Code, to include specific instructions regarding the creation of video games. As reported by Pocket Gamer.biz, the change was first identified by a user on X who noticed a new rule added on October 27th. This update specifically tells the AI tool not to recreate any existing games.

The new guidelines are designed to be firm regarding the nature of the request. The instructions state that the AI should refuse to recreate a specific game even if the user provides a different name for the project. It will also refuse the request if the user describes the project as an original work while it clearly mimics an existing title.

Anthropic staff member Addy Osmani confirmed that this change was added to the Claude Code prompt on Opus 5.5. The company stated that using the tool to infringe on intellectual property goes against its established usage policy. This move represents a formal step toward balancing AI utility with legal and ethical boundaries.

Rising fears of AI cloning

The decision comes at a time of increasing anxiety for independent game developers. Many creators worry that others could feed their existing games into AI tools to generate competing versions. This fear of automated cloning has led some studios to take defensive measures to protect their work.

For example, the studio Digital Sun Games recently made a significant decision regarding their upcoming title, ReVamp. The company cancelled a public playtest for the game because they were concerned that someone might use the demo to create a clone. This highlights how developers are currently weighing the benefits of public exposure against the risks of AI-driven replication.

These concerns are not isolated to a single studio but reflect a broader trend in the industry. Developers are looking for ways to ensure that their unique designs and assets are not easily duplicated by others using generative technology. Anthropic's new restriction is a direct response to these specific industry pressures.

The rise of vibe coding

The growth of AI in game development is visible in different regions, including Africa. A trend known as vibe coding has gained significant traction there following the success of a game called Lagos Life. This web-based simulation game launched on October 1st and has already reached over 60 million visits.

The success of Lagos Life has encouraged more developers to use AI tools to build web games quickly. The parent company of the game, Vatar, even managed to raise $500,000 in funding following the title's unexpected popularity. This demonstrates how AI is lowering the barrier to entry for creating and distributing interactive content.

While vibe coding allows for rapid creation, it also contributes to the conversation about how easily games can be produced. As more people use these tools to build titles, the need for clear boundaries on what can be copied becomes more urgent. Anthropic is attempting to navigate this landscape by allowing genre-based creation while blocking specific copies.

AI's role in the gaming market

The influence of AI on the gaming landscape is becoming increasingly measurable. Roblox reported that more than 80% of its top 100 games by playtime used AI tools in October 2026. This data suggests that AI is already a primary method for accelerating development for many high-traffic titles.

Despite the widespread use of these tools, Anthropic maintains that users can still use Claude to build original games. The tool remains open for creating titles within established genres and using a user's own unique designs and assets. The goal is to support the creative process without facilitating the theft of existing work.

However, the effectiveness of these restrictions may be limited by the variety of models available. While Anthropic has placed these limits on Claude, developers could still turn to open-weight AI models. These models often do not have the same built-in restrictions, which may present a continuing challenge for those trying to prevent AI-generated clones.