Anthropic takes steps to block Claude from reverse-engineering video games

Anthropic’s Claude has reportedly begun refusing or limiting requests involving video-game modding, decompilation, and reverse-engineering. The restrictions appear aimed at preventing AI from helping users bypass copyright protections or reproduce games without permission, though Anthropic has not publicly explained the change. The concern follows a wave of AI-assisted experimentation: modders have altered game code to place characters such as Mario in Elden Ring or Spider-Man in Batman: Arkham City, while others have made browser-playable versions of World of Warcraft, Call of Duty, and Grand Theft Auto: Vice City. A PlayStation 5 jailbreak further enabled older-console compatibility and attempts to run games associated with rival platforms.

The limits may also obstruct legitimate game developers and routine troubleshooting, making the policy broader than simply blocking piracy. The stakes for rights holders include potential lawsuits and substantial damages, but restricting one AI tool may not stop determined users if competing models remain willing to assist. That creates an incentive problem: companies that enforce copyright protections could lose users to less restrictive services. The situation also exposes a contradiction in the AI industry’s stance on intellectual property: companies that built models using vast collections of copyrighted material are now drawing firm boundaries around how users may reuse game code and content.