Over the past month, there has been a sudden surge in browser ports and recompiled games, tasks that previously took years to develop. This is due to recent advancements in AI tools like Claude, which allow even individuals without programming basics to perform software reverse engineering.
Classic PS2 titles like Resident Evil: Outbreak can now run directly in browsers, The Legend of Zelda: Breath of the Wild has received a native PC port, and even the PS5 and Switch 2 consoles have been cracked—the Pandora's box has been thoroughly opened.
It is entirely understandable that passionate hobbyists, who invest significant effort into polishing high-quality results, feel dissatisfied seeing their work drowned out by poorly made "vibe-coded" junk. Meanwhile, publishers are stuck in an endless game of "whack-a-M.O.L.E," trying to prevent players from enjoying titles like Mario Kart World via Google Chrome. More alarmingly, the lowered barrier to software reverse engineering has exposed serious security risks: AI identified a vulnerability hidden in PHP for over a decade within just hours, putting 70% of the global internet at risk and forcing governments to urgently take down public websites to prevent malicious entities from seizing control.
Clearly, all parties are feeling the pressure, with Anthropic and OpenAI now tightening related controls. Anthropic updated its real-time cybersecurity classifier to prevent users from performing automated decompilation operations via the community-developed "Reverse Everythinggj Assembly" (REA) framework; reportedly, OpenAI has also banned relevant accounts citing "online abuse." Since many low-quality recompiled and browser-port titles rely on these tools, the explosive growth of such low-barrier ports may have begun to slow. However, once Pandora's box is opened, it can never be closed again.
Users have begun shifting toward local deployment or open-source alternatives, particularly domestic AI programs, to continue software reverse engineering efforts. Since AI has proven capable of uncovering security vulnerabilities, identifying human oversights, and recompiling programs, its impact extends far beyond the gaming industry. Publishers are certainly frustrated—anyone can now crack consoles and port games to any platform simply by paying for computing power—but the core issue worth noting is that AI agents are attacking public infrastructure faster than patches can fix it.
Claude and OpenAI implemented rules banning online abuse not merely to stop players from running Animal Crossing in Firefox—these AI companies acted urgently because the same techniques expose the fragile nature of our current digital systems. Given the rapid iteration of AI technology, it is not hard to imagine what will be compromised next: smart devices, firewalls, or even public transportation and healthcare systems. After all, AI agents have already successfully infiltrated user bank accounts and stolen hospital data.
Therefore, in addition to causing significant environmental impact and disrupting the job market, AI tools are further exacerbating cybersecurity risks, making our digital world more vulnerable to attacks than ever before.
Furthermore, Anthropic, the developer of Claude, is not limiting its controls to restricting software decompilation. On October 27, Claude quietly added a new rule regarding game remakes that cannot be disabled. The rule states: "Support the development of original games, specific genres, and games based on users' own designs and assets. Recreating specific existing games is prohibited, even if renamed or claimed to be original works."