Anthropic, the company behind the Claude AI tool, has introduced a new rule prohibiting users from recreating specific existing video games, even if they are renamed or rebranded as original. The move comes in response to growing concerns from game developers about their work being cloned or copied using AI tools before their official releases. Spotted by users around October 27, the update explicitly instructs Claude: ‘Do not build a recreation of a specific existing game, even if it is renamed or called original.’
The change was confirmed by Addy Osmani, a member of Anthropic’s technical staff, who explained that the rule aligns with the company’s usage policy against infringing on intellectual property (IP). ‘Original games, genre games, and games built from your own designs and assets are meant to keep working,’ Osmani clarified in a post.
Developer Concerns Drive the Change
The update follows reports of developers fearing their games could be cloned using AI. Alliance Arts, for instance, canceled the public demo of its upcoming game Drunken Goddess Reflux, citing worries that its content could be replicated using AI tools. Other developers have been pressuring platforms like Steam to address the influx of cloned games.
However, the effectiveness of the new rule remains uncertain. Claude lacks the ability to verify whether a user owns the rights to the game they’re working on, and other AI models without such restrictions can still be used for cloning.
Mixed Reactions from the Community
The decision has sparked debate among users. Some argue that the rule is performative and easily circumvented. ‘This seems to stop you from lazily prompting “copy this game but for Roblox,” but it’s trivial to circumvent—you just have another AI write up a game design document based on the original and give that to Claude,’ one user noted.
Others criticized Anthropic’s selective approach to IP protection. ‘So they care selectively about intellectual property? It’s only a problem when it starts affecting the bottom line of Sony or Disney,’ another user quipped.
Some developers expressed frustration over the rule’s impact on legitimate work. ‘It has made Claude completely useless for cybersecurity work. You can’t even prompt for it to help you find flaws in your code without it freaking out and blocking everything,’ one user shared.
Broader Implications for AI and IP
The update highlights the ongoing tension between AI capabilities and intellectual property rights. While Anthropic aims to strike a balance, critics argue that such rules may favor large corporations over individual creators. ‘Big companies’ stance on AI—are we making money? Pro. Is it stealing our IP? Hate it. None of them actually back it other than big tech cos they keep getting government funding,’ one commenter observed.
Anthropic has pledged to continue refining its policies, inviting feedback from users whose projects may be mistakenly flagged. However, the debate over AI’s role in IP infringement is far from settled, leaving both developers and users navigating a rapidly evolving landscape.