Switch 2 Jailbroken: When a Coding Game Becomes a System Backdoor

A few days ago, a strange name appeared on X: Sammy Hajhamid, also known as pepsipu. He posted a short video showing nothing special except a scene from Mario Kart World running on a Nintendo Switch 2 with modified graphics, Peter Griffin appearing instead of Mario. The video was less than thirty seconds long, but it marked the first time a Switch 2 was fully software-jailbroken, without hardware intervention. The tech community buzzed immediately. And just hours later, something stranger happened: SmileBASIC 4, a programming game that had been popular on the eShop, vanished from the digital storefront in multiple regions. To understand what's happening, we need to place this in the broader context of the gaming console industry. Nintendo is famous for its extremely strict security policies. Every new console generation is a challenge to hackers, and the Switch 2 is no exception. The device shipped with firmware 23.0.1, incorporating a host of new protections, from memory encryption to system integrity checks. Yet according to what pepsipu released, the vulnerability wasn't in those protections. It was in a game called SmileBASIC 4 – a miniature programming environment that lets users write code and run simple programs directly on the console. SmileBASIC 4 is not a typical entertainment game. It's a programming sandbox where users can create small applications, mini-games, and even code snippets that interact with the system. As such, it has always been a favorite target for security researchers. But this time, pepsipu didn't just find a bug in SmileBASIC 4. He used artificial intelligence to find it. Specifically, he fed the source code of SmileBASIC 4 into OpenAI Codex, a language model specialized in programming, and asked it to analyze potential weaknesses. Codex identified a buffer overflow in the system file service that SmileBASIC 4 uses to store user data. This flaw allows an attacker to overwrite the service's memory, thereby executing arbitrary code at the system level. Pepsipu took about two weeks to develop the exploit, plus one day to refine and record the demonstration video. The result is a jailbreak called ns2jb, operating via a tethered method – meaning each time the console restarts, the user must connect it to a computer to re-activate. Still, this is a breakthrough: the first time a Switch 2 has been fully software-controlled, without modchips or physical intervention. What's notable is Nintendo's response. No official statement from the company. No new firmware patch released immediately. Instead, SmileBASIC 4 quietly disappeared from the eShop. Users in Japan, South Korea, and some European regions reported being unable to find the game in the digital store. SmileBoom's official website, the developer of SmileBASIC 4, also offered no comment. A suspicious silence. In the gaming industry, removing a game from the eShop is generally understood as a platform protection measure. Nintendo may have determined SmileBASIC 4 was a potential attack vector and decided to remove it before anyone else could exploit it. But this is a dangerous move. Removing a game doesn't stop those who already bought it from continuing to use it. It only prevents new users from accessing it. More importantly, it raises questions about digital ownership: users paid for a product, but that product can vanish at any time for security reasons without prior notice. From a technical standpoint, pepsipu's jailbreak is not a disaster. It's a reminder that any system allowing users to run arbitrary code – even in a sandbox environment – carries inherent risk. SmileBASIC 4 was designed for users to write code, and that very capability opened the door. Nintendo may have underestimated the complexity of this programming environment. But there's another aspect that's received less attention: the role of artificial intelligence in finding the vulnerability. Pepsipu is not a professional black-hat hacker. He's an independent researcher, and the tool that helped him find the bug was OpenAI's Codex. This is the first time a console vulnerability has been discovered with AI assistance, and it sets a worrying precedent for Nintendo. If AI can help find bugs in days, Nintendo's patching speed will never keep up. Of course, caution is warranted. Pepsipu has not released full technical details. What he posted is only a proof of concept. There has been no independent confirmation from other security researchers. And Nintendo, with its experience dealing with numerous jailbreaks in the past, may be preparing a firmware 23.0.1 patch within days. At that point, ns2jb will become useless, and SmileBASIC 4 may return to the eShop with a security update. But even if that happens, this event still leaves a lesson. In a world of increasingly intelligent machines and increasingly powerful AI tools, the line between a game and a security hole becomes ever more fragile. A programming game for children can become the key that opens the entire system. And an independent researcher with a personal computer can do what previously only large hacker groups could achieve. For Nintendo, the question is no longer how to prevent jailbreaks. The question is how to maintain user trust in a world where any software can be an attack vector. And for people like pepsipu, the question is whether they can profit from their discoveries without being targeted legally. Because in this game, the winner is not the one who finds the vulnerability fastest, but the one who knows how to use it at the right moment. In Nintendo's closed boardroom, no one shouts louder than the one who is afraid. And their biggest fear is not a single jailbreak. It's the idea that anyone, anywhere, with an AI tool, could become the next threat.