Anthropic Blocks AI Game Clones in Claude Opus Update

Anthropic Blocks AI Game Clones in Claude Opus Update

The Problem of 'AI Slop' on Steam

The rise of generative AI has led to a surge of low-effort, AI-generated game clones—often referred to as "AI slop"—on major digital storefronts like Steam. This issue came to a head recently during Valve's Steam Next Fest, where players complained that their feeds were flooded with clones of existing games, as well as titles that had only just released demos. This influx poses a direct threat to independent developers who invest months or years into creating original titles. In response to the overwhelming volume of copycat titles, some developers, including Alliance Arts (creator of *Drunken Goddess Reflux*) and the team behind *Who You*, canceled their Steam demo plans in favor of private playtests. Valve itself acknowledged the issue, stating it was "thinking about how to solve this problem."

Anthropic's Technical Intervention

In a move to curb the misuse of its technology, Anthropic has implemented a new system-level rule within its Claude Code tool, specifically on the Opus 5.5 model. According to reports by Game Developer and observations by X/Twitter user Chasm, the update was deployed under an internal flag referred to as "tengu_heron_brook" around late October. Unlike typical user-configurable settings, this rule cannot be disabled by end users.

The directive instructs the AI to focus on original creations while explicitly banning the replication of existing intellectual property. The full text of the rule states: > "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."

When contacted by Game Developer, Anthropic pointed to an explanation by Addy Osmani, a member of the company's technical staff. Osmani confirmed that two sentences were added to Claude Code's prompt on Opus 5.5 to steer the model away from recreating specific games, noting that using Claude to infringe intellectual property violates the company's usage policy. Osmani emphasized that original games, genre-based games, and projects built from a user's own designs and assets are intended to remain fully functional. He added that Anthropic will continue to iterate on the system to strike the right balance and invited developers who experience issues to reach out directly.

Industry-Wide Tensions

The flood of AI-generated content has sparked intense debate within the gaming industry. Epic Games' Tim Sweeney has argued against mandatory disclosures of generative AI usage on storefronts, while renowned game designer Masahiro Sakurai has likened AI-generated games to "low-quality goods." Meanwhile, platforms like VRChat have faced significant backlash over their handling of AI-generated user content. Anthropic's latest update represents one of the first concrete technical guardrails put in place by an AI developer to address these concerns.

Practical Implications and Verification

For game developers and hobbyists, the update means that while Claude can still assist with original game design, genre-based concepts, and asset creation from scratch, it will actively refuse requests to clone or closely mimic specific, existing titles.

It remains unclear whether a specific incident prompted Anthropic to deploy this rule, or if it was a proactive measure in response to the broader "AI slop" trend. Users who believe their legitimate projects have been incorrectly flagged or blocked are encouraged to report the issue directly to Anthropic technical staff, as suggested by Osmani, to help refine the model's understanding of originality.

Key Takeaways of the New Rule

• Originality First: Claude will actively encourage the creation of original games, genre-based games, and projects built from user-owned designs and assets.
• No IP Cloning: The model is programmed to reject requests to recreate specific existing games, even if they are renamed or presented as "original."
• No User Override: The rule is enforced via a server-side flag (reportedly named "tengu_heron_brook") and cannot be turned off by standard users.
• Feedback Channel: Developers who face issues with false positives are invited to contact Anthropic technical staff directly to report the blockage.