Sunday, 25 October 2026Global tech news & guides — what changed, and what it means for you.
AuTechWire
Illustration: a handheld gaming console silhouette on a dark desk at night, its screen glowing with code and circuit traces spilling past the screen edges

Switch 2’s First Jailbreak: What AI Actually Cracked

by

In brief

Last checked October 24, 2026: Nintendo pulled SmileBASIC 4 from the eShop (Oct 23); the tethered ns2jb exploit targets firmware 23.0.1; Nintendo has not otherwise commented.

  • On October 22, 2026, reports from Kotaku and Polygon confirmed the first working jailbreak for the Nintendo Switch 2, roughly 16 months after the console’s June 2025 launch.
  • The person behind it says they co-created the exploit with OpenAI Codex over about two weeks — a fraction of the time console hacks used to take.
  • It is a tethered, in-software exploit: it runs mods and homebrew, but resets every time the console powers off, requires re-running a specific backward-compatible game at each launch, and needs a $25 eShop purchase to work at all — and Nintendo pulled that game (SmileBASIC 4) from the eShop on Oct 23, so new buyers can no longer get it.
  • The exploiter publicly named the eShop game involved, so Nintendo can remove it — and history says the company moves fast and hits hard on piracy.

Last verified: October 23, 2026.

What happened, in order

The jailbreak arrived in two stages. Back in July, researchers found a “userland exploit” — a way to run outside code inside a specific app or game on the Switch 2. That was a real dent in Nintendo’s defenses, but not a jailbreak: it ran inside one app, not across the system.

This week took it a stage further. The person responsible — posting as pepsipu on X — showed footage of Mario Kart World running with Peter Griffin, Hatsune Miku, and Sanrio’s Cinnamoroll modded in. Centro Leaks amplified a video clip on October 22 declaring the first jailbreak released. Kotaku and Polygon both covered it the same day, with Polygon reporting the exploit’s core ingredient: a $25 purchase of the programming app SmileBASIC 4 from the Nintendo eShop.

The individual says the hack took “substantial effort” over about two weeks, working with OpenAI Codex, in a setup one report described as resembling a Rube Goldberg device. Two weeks is the detail worth underlining. Console jailbreaks used to be measured in years of human effort; this one went from userland foothold to working jailbreak on an AI-assisted schedule.

What it actually does — and the three caveats

The exploit lets users run mods, install homebrew software, and download save files — and, inevitably, potentially pirated games. If a console can install a homemade game, it can install a ROM of an official one. That dual use is exactly why the “homebrew” framing that always accompanies these announcements is treated skeptically by platform holders.

But three caveats keep this far from a mass-market jailbreak:

1. It resets when the power goes off. This is a tethered exploit — the modification is not permanent. Every shutdown wipes it, and the entire process has to be run again after every launch. Polygon notes the instructions are “incredibly hard to parse for the average gamer,” so this is not a casual afternoon project.

2. It needs a specific game, every time. The method requires running a particular backward-compatible Switch 1 game to trigger the exploit chain — on every launch. The person responsible publicly identified which eShop game is involved. Nintendo made that counter-move on Oct 23, delisting SmileBASIC 4 from the eShop — the entry point is now closed for anyone who hadn’t already bought it.

3. Nintendo can — and will — patch it. Unlike the original Switch 1, which was cracked through a hardware flaw that could only be fixed in new manufacturing runs (leaving launch units permanently jailbreakable), this exploit is pure software. A firmware update can close it. Kotaku reports Nintendo has been contacted for comment; no response was public as of publication.

Kotaku also notes one genuine advance for the hacking side: unlike the early Switch 2 ROM-cart adapters, which Nintendo detected and used to ban users from their own devices, this exploit is said to evade Nintendo’s online checks. That advantage lasts only until the firmware update lands.

Why the AI part is the real story

Sixteen months is fast for a jailbreak — but the timeline that matters is the one behind it. Console security has always been a war of attrition: humans probe firmware for months or years, find one foothold, and the manufacturer patches it. AI coding tools compress that loop. The exploiter credited OpenAI Codex directly, and Polygon places the Switch 2 hack inside a wider wave: the PlayStation 5 has been receiving firmware updates specifically to fight new jailbreaks popping up faster than ever, with Polygon noting the pressure building ahead of Grand Theft Auto 6.

The same acceleration is visible across gaming’s homebrew scene — browser ports of the original Halo: Combat Evolved, viral videos of Mario modded into Elden Ring and skateboarding added to Call of Duty — all attributed by Polygon to AI coding, particularly following the September release of Claude Opus 5.5. Even the platform holders are reading the room: Xbox CEO Asha Sharma recently announced Xbox Labs, a division that — while never naming AI directly — appears heavily inspired by the vibe-coding trend.

This is the same arc this publication has been tracking all month: AI compressed the find-to-exploit window on PHP vulnerabilities, armed the CARLIY0S bank-heist ring, ran the ConnectWise supply-chain campaign through the ARTEX pentest tool. Console jailbreaks are now part of that pattern — offense at machine speed, across every layer of the stack.

What Nintendo does next

History gives a good read on the company’s playbook. When Switch 1 units were being sold pre-jailbroken, Nintendo came down on the sellers hard. When ROM-cart adapters appeared for the Switch 2, the company detected the usage and banned the users from their own devices. A software exploit with a publicly identified entry game gives Nintendo three fast moves: delist the game, push a firmware patch, and flag accounts running the modified software. All three are cheap for Nintendo and expensive for the jailbreak to survive.

The realistic outcome: the window for this exploit is measured in weeks, maybe days. The longer-term question is whether AI-assisted probing keeps producing new entry points as fast as Nintendo closes them — turning console security from a once-a-generation hardware battle into a permanent, fast-moving patch treadmill.

What this means for you

  • If you own a Switch 2: don’t install anything from this wave. Nintendo has a track record of banning modified consoles from online services, and a tethered exploit that resets on every reboot is not worth the risk of losing your account and game library.
  • If you are considering buying a Switch 2 this holiday season: this jailbreak changes nothing about what you should buy. It is not stable, not permanent, and not user-friendly — and the $499 street price (up from the $449.99 launch MSRP) matters more to your wallet than an exploit that will be patched.
  • If you follow game security: Nintendo delisted SmileBASIC 4 on Oct 23 — watch how fast the firmware response lands. The speed of that response — not the jailbreak itself — is the real measurement of where the AI-offense era has left console security.
  • If you make games: the cost of cracking your platform’s protections just fell dramatically. Protections assumed to buy years now buy months. Plan accordingly.

Sources

Shoaib Hassan

Shoaib Hassan is the founder and editor of AuTechWire, an independent technology publication covering broadband, mobile networks, privacy, startups, and consumer tech.

Related Stories

Leave a Comment