Anthropic has updated its AI coding assistant, Claude Code, to prevent the creation of clones based on existing video games. The modification was first identified by X user Chasm, who noted that the change was deployed on October 27. Operating under the internal designation "tengu_heron_brook," the new rule instructs the model to generate original games, titles within specific genres, or projects based on user-provided designs and assets. Crucially, it explicitly prohibits recreating a specific existing game, even if the title is renamed or claimed to be original. According to Chasm, users do not have the option to disable this restriction.
Addy Osmani, a member of technical staff at Anthropic, later confirmed the update on X. Osmani explained that two sentences were added to the prompt for Opus 5.5 to steer the model away from replicating someone else's specific game, noting that using Claude to infringe on intellectual property violates the company's usage policy. Osmani emphasized that the tool remains functional for creating original games, genre-based projects, and titles built from a user's own designs and assets, adding that the team will continue to iterate to ensure the right balance is struck.
Despite these measures, questions remain regarding the effectiveness of the rule. As noted by Game Developer, safety restrictions on AI models can often be bypassed through carefully worded prompts. Furthermore, there is no clear mechanism for Claude to verify whether a user actually owns the game designs or assets they claim to be using. This update arrives amid growing concerns among developers about the misuse of AI tools to clone games. Recently, Digital Sun delayed a public playtest of its title ReVamp due to fears that its demo could be fed into an AI tool, while Valve has faced pressure to explain how it will prevent AI-made clones from crowding out original games on Steam.
Sources: GamesIndustry.biz.