Claude Code now refuses to clone specific video games, a small IP guardrail that won't stop determined copycats

🕒 Published on Zendoric: November 3, 2026 · 00:06
✨ AI-generated · how it's made
Anthropic quietly added two sentences to Claude Code's prompt that block recreations of specific existing games. The company is reacting to a wave of AI-made clones that has frightened indie studios. The rule sends the right signal, but it can only do so much.
Anthropic has changed Claude Code so it will not recreate existing video games. X user Chasm spotted the new rule, pushed around October 27 under the internal label 'tengu_heron_brook'. It reads: 'Do build original games, games in a genre, and games from the user's own designs and assets. Do not build a recreation of a specific existing game, even if it is renamed or called original.' According to Chasm, users cannot turn it off.
An Anthropic spokesperson pointed Game Developer to a post by staff member Addy Osmani. He confirmed the company 'added two sentences to Claude Code's prompt on Opus 5.5' to steer it away from copying someone else's game, because 'using Claude to infringe IP is against our usage policy.' He said original and genre games should keep working, that Anthropic will 'continue to iterate,' and that developers whose own projects get refused can send them to him.
The context explains the move. For months, indie developers have watched AI-made remakes of classic and upcoming games appear, some made for fun and some for money. Digital Sun Games is one of several studios that cancelled a public playtest for its game ReVamp. Its fear was that someone could feed the demo into an AI tool and release a clone first. Developers have also pressed Valve on how Steam will stop such clones, and Valve gave South Korean developers a rare explanation this week. In the same news cycle, Game Developer reported that the number of game pages on Itch.io has doubled as AI games flood the store.
The rule has clear limits. Researchers and reporters have repeatedly shown that rhetorical tricks can get around safety limits in generative AI tools. It is also unclear how a model would decide who owns a game idea, or where a genre game ends and a copy begins. A prompt instruction is a soft barrier. It stops the laziest copycats, not determined ones, and it may sometimes block legitimate projects.
Our reading: this matters less as a technical fix than as a precedent. A model provider is admitting that what its agent can build carries responsibility, and is acting before a court or regulator forces it to. That fits a pattern we have been tracking: agentic coding makes production almost free, and the bottleneck moves to legitimacy, meaning what should be built and for whom. Over time, the answer cannot rest on two sentences in a system prompt. It will need provenance signals, store-level policing and clearer IP norms for derivative work. The upside is still large. Tools that let a solo developer build an original game in weeks are a real expansion of who gets to create. The goal is to keep that while making cloning unprofitable, not merely discouraged.
🔗 Related on Zendoric
- Anthropic curbs video game clones in Claude Code: a useful rule, but easy to get around · 2026-10-31
- Claude Code refuses to clone specific video games: a useful patch, but the line between inspiration and imitation remains undrawn · 2026-11-01
- Anthropic stops Claude Code from cloning other companies' games: a legal precaution with an IPO in sight · 2026-11-02


