Anthropic appears to block Claude from cloning games

Anthropic appears to have narrowed what its Claude assistant will build when asked to make games. According to the latest report, a recent update seems to stop Claude from producing outright clones or copies of existing titles. Instead, the model is now pushed toward making what it treats as original games.

The change arrives in the middle of an ongoing argument over AI-generated copies of games. Developers and players have raised concerns that generative tools can reproduce well-known titles with little effort, which raises questions about copyright, credit and the value of original design work. Those worries have grown as AI systems get better at writing working code and assembling simple interactive projects.

Guardrails like this are a familiar move for AI companies. Model makers often add limits around requests that could infringe intellectual property, but the line between a homage, a remix and a copy is hard to draw. A filter that refuses obvious clones may still allow close imitations, and the reporting available does not say how strictly the rule is applied or whether rewording a request gets around it.

For independent developers, the question is practical. Many small studios and hobbyists use AI assistants to prototype mechanics, generate starter code and test ideas quickly. A restriction on copies would not stop that work, but it would limit one shortcut some people have used to recreate existing games.

The picture is still incomplete. The change is described as seemingly in place, which suggests it was noticed by users or observers rather than laid out in full detail. Until Anthropic clarifies what counts as a clone and how the update works, the scope of the restriction should be treated as provisional.