HomeAIAI Policy & SafetyAnthropic Steers Claude Code Away From Cloning Existing Games

Anthropic Steers Claude Code Away From Cloning Existing Games

Anthropic has told Claude Code to stop recreating other people’s games. Two new sentences in the coding tool’s prompt on Opus 5.5 steer it away from building a copy of a specific existing title, while still allowing original games, genre games and games made from a user’s own designs.

The change came to light when an X user called Chasm posted the new rule on 30 October. Later that day, Addy Osmani, whose X profile describes him as working on “Claude Code @anthropicai”, confirmed it in a reply. “We added two sentences to Claude Code’s prompt on Opus 5.5 to steer it away from recreating someone else’s specific game because using Claude to infringe IP is against our usage policy,” he wrote.

What the new Claude Code rule says

According to Chasm, the rule first appeared around 27 October and sits under an internal flag named “tengu_heron_brook”. The text Chasm posted reads: “Do build original games, games in a genre, and games from the user’s own designs and assets. Don’t build a recreation of a specific existing game, even if it is renamed or called original.” Chasm said there was no setting to switch it off. TSN has not inspected the prompt itself.

The last clause is the important one. Renaming a copy, or simply calling it original, is not meant to get it past the rule. That leaves Claude Code to judge whether a request describes a new game in a familiar genre or a near-exact copy of one that already exists, a line that human designers and courts also find hard to draw.

Osmani set out what should still work. “Original games, genre games, and games built from your own designs and assets are meant to keep working,” he wrote. “We’ll continue to iterate to ensure we’re striking the right balance. In the meantime, if Claude turns down something that’s yours, send it my way.”

Why Anthropic points to its usage policy

Osmani tied the change to Anthropic’s rules, not to a new law or a court case. Anthropic’s Usage Policy, in the version published on its site that takes effect on 12 November, covers its apps “such as Claude.ai and Claude Code”. It bars using them to “infringe, misappropriate, or violate the intellectual property rights of a third party”.

The same policy makes a point that matters for developers whose requests get refused. “Our products also include real-time safeguards that may block or limit certain outputs, but encountering a block does not by itself indicate a violation of the Usage Policy,” it says. In other words, a refusal from the new rule is not an accusation.

The model in question is Claude Opus 5.5, whose fast mode left research preview earlier in October. PC Gamer, reporting on the change via Game Developer, said Opus 5.5 has proved adept at reverse-engineering code, and that several studios had cancelled public demos for fear their code would be copied into AI clones.

Why a prompt rule will not settle the clone problem

The pressure on Anthropic comes from game makers, not regulators. According to PC Gamer, Steam has been flooded with low-effort, vibe-coded games, many of them close imitations of existing titles, and the latest Steam Next Fest had twice as many demos as the one before, with AI-assisted games making up most of the increase. Valve’s Tom Giardino told developers at a Steam event that Valve does not want “to be gatekeepers telling developers which tools they should or shouldn’t use”, PC Gamer reported.

A rule in one tool’s prompt is a narrow answer to that. It applies to Claude Code on Opus 5.5, by Osmani’s account, and says nothing about other models or other coding assistants. Anyone determined to copy a game can still describe its mechanics piece by piece, and the rule depends on the model recognising a “specific existing game” from a description. The risk runs the other way too: an indie developer rebuilding their own game, or making something in a crowded genre, may be refused, which is why Osmani asked people to send him such cases.

Still, it is a notable step. AI companies usually point to their usage policies after the fact. Here, Anthropic has written a copyright concern directly into the instructions its coding agent follows, and has done so in the open after a user found it. Game studios have been among the first to feel what cheap AI coding can do, as Epic’s own vibe-coded games launcher showed from the other side. Claude Code’s popularity has drawn other kinds of abuse too: TSN recently reported how a tampered Claude Code extension reached up to 5,000 users. Wider legal fights over AI and creative work are tracked on TSN’s AI copyright hub.

What to watch

The next signs will be how often developers report refusals for their own work, and whether Anthropic changes the two sentences, as Osmani said it will keep iterating. Its updated Usage Policy takes effect on 12 November.

Sources

Related stories

Image: PattayaPatrol / Wikimedia Commons, CC BY-SA 4.0, modified by TSN; this image is also CC BY-SA 4.0

Follow @tsnmediahub on X for breaking AI news as it happens.

Follow on X

Share this story

Latest stories

More in this category

Latest stories

Explore the TSN Atlas: Every topic we cover, mapped. Click a bubble to dive in.