Anthropic and OpenAI Tighten AI Decompilation Controls, Raising the Bar for Browser Ports and Game Remakes

Classic PS2 games like Resident Evil: Outbreak can now run directly in a browser, a native PC port of The Legend of Zelda: Breath of the Wild has appeared, and even the PS5 and Switch 2 consoles have been cracked one after another—Pandora's box has been flung wide open.

It is entirely understandable that hobbyists who truly love these games and pour enormous effort into polishing high-quality results feel dissatisfied as their work is drowned out by slapdash "vibe-coded" garbage content. At the same time, publishers are caught in an endless game of whack-a-mole, busy stopping players from playing titles like Mario Kart World through Google Chrome. Even more alarming, the lowered barrier to software reverse engineering has exposed serious security risks: AI took only a few hours to find a vulnerability that had lain hidden in PHP for more than a decade, putting seventy percent of the world's networks at security risk, and governments even took public-facing websites offline in emergencies to prevent malicious actors from seizing control.

Clearly all parties are feeling the pressure, and Anthropic and OpenAI have now begun tightening related controls. Anthropic updated its real-time cybersecurity classifier to block users from performing automated decompilation through the community-developed "Reverse Engineering Anything (REA)" framework; OpenAI reportedly also banned related accounts on grounds of "cyber abuse." Because a large number of low-quality recompilations and browser ports rely on such tools, the explosive growth of this low-barrier port content may already be starting to slow. But once Pandora's box is opened, it can never be closed again.

Users have already begun turning to local deployments or open-source alternatives, especially domestic Chinese AI programs, to continue software reverse engineering work. Since AI has been proven capable of uncovering security vulnerabilities, finding human oversights, and recompiling programs, its impact has long since gone far beyond gaming. Publishers will certainly be annoyed—anyone who can afford to buy computing power can crack consoles and port games to any platform—but the core issue truly worth attention is this: AI agents are attacking public infrastructure faster than patches can be fixed.

Claude and OpenAI's cyber-abuse ban rules are not just about stopping players from playing Animal Crossing in Firefox—these AI companies acted urgently because the same technical means have exposed the fragile nature of our existing digital systems. Given the pace of AI iteration, it is not hard to imagine what field will be breached next: smart devices, firewalls, and even public transportation and healthcare systems. After all, AI agents have already successfully broken into users' bank accounts and stolen hospital data.

In addition, the controls of Claude's developer Anthropic are not limited to restricting software decompilation. On October 27, Claude quietly added a rule targeting game remakes, and the rule cannot be turned off. The rule states: "Support the development of original games, games in specific genres, and games based on the user's own designs and assets. Reproducing specific existing games is prohibited, even if the name is changed or it is claimed to be an original work."