Anthropic Adds 2 Sentences to Claude Code to Stop AI Video Game Clones
Anthropic has added two sentences to Claude Code that cease its Opus 5.5 mannequin from cloning current video video games. Developers say the filter additionally blocks crash dump evaluation, anti-cheat work, and debugging of their very own tasks.
The guardrail targets a wave of AI-built sport clones. It additionally arrives earlier than Anthropic’s Nasdaq debut. The firm already paid $1.5 billion to settle claims over pirated coaching books.
What Exactly Is Claude Now Refusing?
X consumer Chasm noticed the instruction on October 27. Original video games, style titles, and tasks from a consumer’s personal property stay allowed.
However, recreating a particular current sport is now off limits, even below a brand new title.
Addy Osmani, a member of Anthropic’s technical workers, confirmed the change on X on October 30. He cited a utilization coverage banning mental property (IP) infringement.
“We added two sentences to Claude Code’s immediate on Opus 5.5 to steer it away from recreating another person’s particular sport.”
Why Game Studios Fear AI Copies
Modders utilizing AI have dropped Mario into Elden Ring and ported Vice City to an online browser.
An AI-aided PlayStation 5 jailbreak even ran Microsoft’s Halo 2 on Sony {hardware}, Mashable reported.
Smaller studios face a extra direct menace. Some builders say scammers threatened to clone their demos into full video games earlier than the originals launch.
Digital Sun Games, for example, cancelled a public ReVamp playtest over copycat fears.
Can Claude Tell an Owner From a Copycat?
Developers say the filter reaches nicely past clones. Claude Code’s GitHub tracker exhibits refusals in the midst of strange sport tasks.
Other reports cite blocked anti-cheat improvement, safety opinions, and even a fundamental Windows crash dump.
Anthropic has not defined how the mannequin decides who owns a mission. Osmani promised additional tuning and requested builders to flag refusals on their very own work.
Why Accept a Developer Backlash Now?
Anthropic’s IPO prospectus, filed October 26, flags legal responsibility from third-party misuse as a threat. It additionally books the authors’ settlement as a 2025 expense.
The timing issues, since buyers are ready on Anthropic’s IPO pricing earlier than valuing rival AI labs.
Meanwhile, builders on Reddit and X say they may swap to Chinese, open-weight, or native fashions. Cheaper rivals are already competing on worth, together with OpenAI’s GPT-6.1 Terra.
Mashable summed up the trade-off going through each lab.
“…the least moral AI is perhaps poised to develop into the most well-liked.”
The put up Anthropic Adds 2 Sentences to Claude Code to Stop AI Video Game Clones appeared first on BeInCrypto.