The PlayStation 3’s arcane architecture is why many never thought we’d actually see Metal Gear Solid 4 on PC, soon to finally be free of its last-last-gen prison. It’s fair to say that PlayStation 3 emulation has come along in leaps and bounds over the years, but the advancement of AI may end up pumping the brakes.

The team behind open-source PlayStation 3 emulator RPCS3 has apparently been inundated with AI code pull requests, recently issuing a (mostly) polite request to contributors via X to “Please stop submitting AI slop code” (via Respawn First). The team warns that those who continue to submit AI-generated code to the project’s GitHub without proper disclosure will be banned.

The team goes on to write, “There are plenty of resources online to learn how to debug and code instead of generating slop that you don’t understand and that doesn’t work.”

The AI guidelines on the GitHub project have been updated. It’s important to note that AI tools are not banned outright, though each pull request must state “the scope of AI involvement.” The guidelines seemingly acknowledge the difference between AI-assisted code and ‘AI slop code,’ stating that AI tools are permitted for “research and reverse engineering purposes,” but there is the expectation that contributors “fully own and understand all code they submit.”

Basically, vibe coders can still generate the code, but will need to then make the effort to comb through the LLM’s output and ensure they’re on the same page as the bot.

The majority of the emulator’s AI code pull requests have apparently been for macOS builds, (only one developer on the main team has the Apple hardware necessary for maintenance, so that’s a lot of pull requests to sift through solo). On this note, the team shares, “We have had to revert a few slop [pull requests] that caused big regressions a few times—enough is enough.”

Proponents will claim AI coding tools represent big wins in terms of efficiency, but often those gains appear to be undone in the implementation. The GitHub’s guidelines echo this point, saying, “[AI slop code] wastes maintainer time and, in worse cases, such changes get merged and break functionality for all users.”

The RPCS3 team aren’t the only devs losing time to dodgy AI code. Just for one example, Linus Torvalds recently lamented that “The continued flood of AI reports has basically made the [Linux team’s] security list almost entirely unmanageable.”

My Dad used to say that on the internet, no one knows you’re a dog—and it turns out the same is true of vibe coding. Doggone code may be the least of a maintainer’s worries, too; one human software engineer rejected an AI agent’s code change request, only to later discover the AI agent had published an ‘angry’ blog about him. So, while vibe coding and coding agents may be quick and easy options, I’m not yet convinced they’re actually the best option.