PS5 Linux lead quits: "a bunch of noobs using LLMs" that "they don't understand"
A lead developer for the PS5 Linux project has quit after an undisclosed hypervisor exploit his work relied on was independently rediscovered with large language models and reported to Sony for a bug bounty, likely leading to the flaw being patched. Commenters debate whether using AI to find and cash in on such vulnerabilities is legitimate security work or a “slop” shortcut that undermines long-running homebrew efforts and community norms. More broadly, the thread reflects growing tension in open source and hobbyist scenes as LLMs flood projects with low-effort contributions, erode traditional skill-based status, and push maintainers toward stricter gatekeeping.
Context: PS5 Linux, Hypervisor Exploit, and Resignation
- Thread agrees the headline overemphasizes LLMs: a central trigger was an exploit disclosure that threatens the PS5 Linux jailbreak path, not just “AI slop.”
- PS5 Linux relied on a hypervisor bug to bypass Sony’s protections. Multiple people (at least three) independently found this bug; one reported it to Sony for a bounty.
- The PS5 Linux dev had also found the bug, kept it secret to preserve Linux support, and says they asked others to delay disclosure until after GTA 6’s release so users could both buy the game and run Linux.
- Some say this disclosure “kills” or “jeopardizes” the project; others argue it only affects newer firmware and older consoles remain exploitable.
LLMs, “Slop Kiddies,” and Bug Bounties
- One camp views the bug-hunter as selfish: using LLMs to find a rare hypervisor bug, ignoring the community’s request, and cashing in while ruining a long-running hobby project.
- Another camp says they followed the rules of the bounty program, had no obligation to sacrifice money for the scene, and that Sony or others could have found the bug with the same tools anyway.
- There is debate whether frustration is really about ethics and community norms versus resentment that newcomers with LLMs can match or beat expert work.
Gatekeeping, Community Quality, and AI-Driven “Vibecoding”
- Many commenters see this as part of a larger shift: open hobbyist/security scenes once filtered by deep expertise now flooded by “noobs” using LLMs without understanding.
- Some defend gatekeeping as necessary to preserve quality and culture; others call this elitist but concede that uncontrolled openness often degrades communities.
- Several report similar experiences in open source: LLM-generated PR spam, low-effort “vibe-coded” contributions, and maintainers burning out.
- Suggested responses include tighter contributor restrictions, AI-disclosure policies, automated triage, and private/inner circles for serious contributors.
Broader Reflections on AI and Programming Culture
- Threads compare this to earlier “JavaScript framework of the week” churn and to “Eternal September” for programming.
- Opinions split: some see LLMs as empowering non-experts and inevitable; others see them as hollowing out the joy of understanding systems and eroding the status and satisfaction of expertise.
- Several predict more closed, curated projects and parallel “slop-coded” ecosystems moving fast but with questionable maintainability.