Anthropic’s AI tool Claude is reportedly refusing requests to recreate existing video games, even when users change the game’s name or describe it as an original project.
Developers have also complained that the restrictions are affecting legitimate work, including software testing, reverse engineering and troubleshooting.
The complaints involve Claude Code and its Opus 5.5 model. Several users have reported refusals over the past week, but Anthropic has not publicly explained the changes.
Claude Is Blocking Video Game Copies
According to user reports, Claude is now more likely to reject requests to recreate a specific existing game.
Requests to build original games or develop titles within a particular genre should still be allowed. However, changing the name of an existing game does not necessarily avoid the restrictions.
Complaints have appeared on Claude Code’s GitHub page, with developers reporting that the tool sometimes refuses requests during development.
Developers Say Normal Work Is Also Affected
Some users say Claude is rejecting tasks that have nothing to do with copying games.
Reported examples include analysing Windows crash reports, removing unwanted software from Samsung TVs and using reverse-engineering tools to understand existing software.
Some developers have also reported refusals when working on anti-cheat systems and security testing for their own games.
Reverse engineering involves examining existing software to understand how it works, identify problems or improve security. It is commonly used in software development and cybersecurity.
However, users disagree on how broadly the new restrictions apply. Some believe the changes mainly target direct copies of existing games, while others say legitimate technical work is also being blocked.
Why Anthropic May Be Tightening Restrictions
One possible reason is copyright and legal risk. AI-generated copies of popular games and software could create problems for developers and technology companies.
The restrictions also come after Anthropic introduced its Cyber Verification Program on October 6. The programme limits access to certain advanced cybersecurity capabilities to approved organisations.
However, it is not clear whether the latest complaints about game development and reverse engineering are directly connected to that programme.
Other AI companies have also faced similar concerns over how their tools are used to reproduce or modify existing software.
What This Means for Claude Users
People using Claude to build original games with their own designs and assets should still be able to work on those projects.
But requests to recreate a famous game, even under a different name, may now be rejected.
Developers, students and freelancers could also face problems if the restrictions affect debugging, security reviews or reverse engineering.
For paying users, the changes could mean that some tasks they previously handled with Claude Code may no longer work as expected.
Anthropic has yet to clarify the scope of the restrictions or explain whether it plans to address complaints from developers.
The changes have also sparked debate over copyright. Critics argue that AI companies trained their models on large amounts of online content but are now placing stricter limits on what users can create with them.
