Anthropic's Claude Now Refuses Requests to Recreate Existing Video Games

Anthropic's AI model Claude is reportedly rejecting user requests to recreate existing video games, even when the requested copy is renamed or described as an original creation. The refusals have been documented by multiple developers using Claude Code, the company's AI-assisted coding tool, and have surfaced in a series of complaints on the tool's GitHub page and across social media.

According to an account shared on X, the tool now follows a policy that permits building original games, titles in a particular genre, or projects based on a user's own designs and assets, while declining to recreate a specific existing game regardless of how the request is framed. Anthropic has not issued an official statement confirming the rule text or the scope of the restrictions.

In practice, general requests such as asking for a racing game or a platformer-style title appear to still be accepted. What is now being blocked is the reconstruction of a specific, well-known commercial game under a different name. Several developers reported that their requests were refused midway through development, while attempting to replicate a game mechanic, or while patching a tabletop-style project.

A broader concern raised by users is that the safeguards are catching legitimate technical work. Developers on Reddit and other platforms say Claude declined tasks such as analyzing a Windows crash dump, removing unwanted software from a Samsung TV, and using a widely known reverse-engineering tool. Some self-identified game developers also reported refusals on anti-cheat development and security reviews for their own titles.

The timing may be connected to Anthropic's Cyber Verification Program, introduced on October 6, which restricts access to the advanced cybersecurity capabilities of its frontier models to organizations covered by a separate initiative. However, the refusal reports only began appearing roughly three weeks after that program took effect, leaving the exact cause unclear.

Not all observers accept the reverse-engineering complaints. One commenter argued that the change only targets straightforward cloning requests and that modding and decompilation work remains permitted. The likely driver is legal risk: AI vendors face potential litigation from software companies and game studios if their tools are used to copy or crack commercial products.

The issue is also affecting smaller developers. Steam Next Fest recently drew criticism over a volume of AI-generated games that appeared derivative of existing titles, and some studios have reportedly been targeted by extortion schemes threatening to turn their game demos into full clones before release. Anthropic is not alone in tightening controls; OpenAI recently deactivated the account of a user who was attempting to clean up their television's software.