Recently, Anthropic released a version update for its AI assistant Claude, formally introducing code restrictions that prohibit users from using the model to recreate existing commercial video games.
According to update details exposed on X, the system instructions explicitly require Claude to create only original games, games of specific genres, and games that use the user's own designs and assets. It strictly prohibits generating remakes of any existing games, even if they are renamed or disguised as original projects.
Anthropic engineer Addy Osmani later confirmed this adjustment through public channels and noted that the move is intended to avoid intellectual property infringement risks and ensure that platform operations comply with relevant usage policies. Currently, the industry is seeing a wave of using large models to decompile and port AAA classic games. Well-known titles such as Halo: Combat Evolved, Call of Duty: Modern Warfare 2, and Grand Theft Auto V frequently appear as fan-made versions on platforms such as browsers. Against this backdrop, Anthropic's copyright risk-control action appears especially urgent.
As an AI company that has previously faced litigation over intellectual property issues and paid a huge settlement, its recent cautious attitude toward copyright compliance and content moderation reflects how the entire AI industry is accelerating the establishment of safety guardrails in response to widespread copyright infringement risks. This will inevitably have a profound impact on the boundaries of future AI-assisted content creation.
