Anthropic to block AI-generated clones of existing video games
The AI company is updating its safety guidelines to prevent users from using Claude to recreate specific copyrighted titles.
The short version
- Anthropic added two sentences to Claude Code under Opus 5.5 to prevent users from rebuilding specific existing games.
- The new rule explicitly forbids recreations of specific games, even if they are renamed or labeled as original.
- Developers are still permitted to use the AI to create original games, genres, or designs from their own assets.
- The move responds to a wave of AI-generated clones and 'slop' titles affecting both AAA and indie gaming.
- The AI model Opus 5.5 was identified as being particularly capable of decompiling games to create browser versions.
Why it matters
This represents a shift in how AI companies manage intellectual property rights in the creative arts. For developers, it means a clearer boundary between using AI as a tool for innovation and using it to bypass copyright laws.
New restrictions for Claude
Anthropic is implementing new internal rules to prevent its AI, Claude, from being used to create copies of existing video games. The company discovered these adjustments within its internal guidelines, which aim to curb the production of unauthorized clones. This update follows a growing trend of users attempting to replicate popular titles using generative AI tools.
A technical employee at Anthropic, Addy Osmani, confirmed the change by explaining that the company added specific instructions to the Claude Code prompt under the Opus 5.5 model. These instructions are designed to steer the AI away from rebuilding specific games owned by others. Osmani noted that using Claude to violate intellectual property is a breach of the company's usage policies.
The new policy provides a clear distinction between creative assistance and copyright infringement. Users are still encouraged to use the AI to build original games, explore specific genres, or develop projects based on their own unique designs and assets. The goal is to ensure the technology remains a tool for creation rather than a shortcut for plagiarism.
The rise of AI game clones
The decision comes as the gaming industry faces a surge of AI-generated clones that affect both large AAA titles and smaller indie projects. Anthropic's Opus 5.5 model has been noted for its ability to decompile games, which allows private developers to create browser-based versions of famous titles. These projects have included recreations of popular franchises like Halo: Combat Evolved and Grand Theft Auto V.
This trend has created significant friction for independent developers who are hesitant to share early demos of their work. There is a growing fear that 'vibecoders'—individuals using AI to generate code quickly—will use these demos as templates. These users might then release low-quality versions, often referred to as 'slop,' before the original developer can officially launch their game.
Beyond simple clones, the technology is being used to create unauthorized PC ports for games that were originally exclusive to consoles. These projects have seen a massive acceleration in production speed due to AI assistance. While some of these ports are technically impressive, they often exist in a legal gray area regarding the rights of the original publishers.
Impact on the indie scene
The indie gaming space is currently struggling with an influx of AI-generated content that threatens to drown out high-quality projects. Data from the Steam Next Fest in October showed that nearly 7,000 demos were submitted, which is roughly double the amount submitted in June. This surge highlights how easily AI can be used to flood the market with numerous titles.
Many of these new submissions are described as 'slop' titles, which are low-effort, AI-generated imitations that lack polish. These projects compete for the same limited attention and public visibility as ambitious, hand-crafted indie games. This makes it increasingly difficult for creative developers to stand out in a crowded digital marketplace.
The sheer volume of these unactioned 'feasibility studies' makes it harder for players to find high-quality content. While the technology allows for rapid prototyping, it also risks devaluing the work of human creators. Anthropic's new rules are a direct response to the difficulty of balancing these creative freedoms with the need to protect existing works.
Finding the right balance
Anthropic is currently working to find the correct balance between blocking copyright infringement and allowing for legitimate creative use. The company wants to ensure that developers can still use Claude to build unique experiences without the AI facilitating the theft of others' work. This requires a nuanced approach to identifying what constitutes a 'recreation' versus an 'original' design.
To manage this, the company is looking for feedback on potential 'overblocking.' If the AI becomes too restrictive and prevents users from making original games that happen to share some common tropes, the company wants to know. The goal is to maintain a helpful environment for developers while strictly enforcing the prohibition on building specific, existing games.
As the gaming industry continues to grapple with the implications of AI, these policy changes will be closely watched. The success of these rules will depend on how effectively the AI can distinguish between a game inspired by a genre and a game that is a direct copy of a specific title. For now, the focus remains on protecting intellectual property from being bypassed by automated code generation.
What to watch
- How Anthropic defines the line between 'genre' and 'specific game' in its prompts.
- Whether other AI companies follow Anthropic's lead in blocking game clones.
- The long-term impact of 'slop' titles on indie game discoverability on Steam.
Try it
Check the Steam Next Fest submission history to see the recent surge in demo submissions.
Sources and links
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.
Listen to the AI Wire headlines
Listen to the AI WireListen to our daily AI shows every morning: on your commute, in the shower, wherever you are
Listen to the daily showsFirst reported by ComputerBase · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
