I saw Mozilla’s Betrayal of Open Source: Google’s Gemini AI is Overwriting Volunteer Work on Support Mozilla in my Mastodon feed and asked Perplexity if there are other critiques. Well, there are.

​

Critiques about SUMO/Gemini translations

Several community threads and reports echo the article you shared, focusing on Support Mozilla (SUMO) and the Kitsune-based translation bot:

• A detailed Reddit discussion describes how the AI translation tool overwrites previously human‑translated support articles, often with lower‑quality text and without any way for volunteers to prevent or permanently correct these changes.reddit​

• Coverage on Slashdot and Linuxiac reports that the long‑standing Japanese SUMO team disbanded after the “Sumobot” started automatically editing and approving Japanese knowledge base articles, erasing community‑approved work and disregarding local translation guidelines.tech.slashdot+1​

• A YouTube explainer video focuses specifically on “Mozilla’s Japanese SUMO Community Quits Over AI,” highlighting how machine translations were auto‑approved within a short window, effectively sidelining human reviewers.youtube​linuxiac​

These sources repeat the same core concerns as the blog post: volunteer work being overwritten, lack of effective human control, and damage to long‑standing localization communities.linuxiac+2​youtube​

Critiques of closed, proprietary Gemini

Other commentary picks up on the “betrayal of open source” framing:

• A video and social posts that directly reference the same “Mozilla’s betrayal” narrative stress that the SUMO bot is built on Google’s closed‑source Gemini 2.5 Pro and argue that this clashes with Mozilla’s public rhetoric about open, community‑driven AI and datasets.instagram+1​youtube​

• Posts on LinkedIn and Mastodon frame this as Google’s Gemini undermining volunteer efforts and as Mozilla abandoning open‑source values by relying on a proprietary model trained on opaque or potentially infringing data.linkedin+1​

These pieces parallel the article’s argument that Mozilla is using a closed, contested model to overwrite an open, community‑maintained corpus, thereby devaluing both open data and contributor labor.youtube​linkedin​

Broader backlash to Mozilla’s AI direction

Beyond SUMO, there is more general criticism of Mozilla’s current AI push:

• Mozilla’s new AI Window and AI services experiments in Firefox have drawn user backlash on Mozilla’s own forums and on Hacker News; critics argue that Mozilla is chasing AI features users did not request, risking privacy and diluting its traditional open‑web mission.tech.yahoo+2​

• Earlier in 2025 Mozilla also faced criticism over new Firefox terms of use, which some users read as overly broad and potentially enabling data use for AI, prompting Mozilla to issue public reassurances.techcrunch​

Other cases of open source projects replacing volunteers with AI​

Translation and localization communities

Outside Mozilla, translation volunteers and projects are debating or resisting AI replacements rather than embracing them wholesale.

• The Open Language Initiative (OLI) has public arguments for why AI tools should not replace its volunteer translators, stressing quality, cultural nuance, and the value of community labor; the project discusses AI more as a supplement than a substitute, precisely to avoid demotivating volunteers.olangi​

• Professional and community translators also report AI being used to automate routine translation tasks, which reshapes roles and can crowd out human work, but this is more visible in commercial settings than in structured FOSS localization programs so far.linkedin​

Code contributions and “AI slop”

In software development, the main pattern is not “projects replacing volunteers with AI,” but volunteers being drowned out or sidelined by AI‑generated content.

• Open source maintainers increasingly report being overwhelmed by AI‑generated pull requests and bug reports, often low‑quality, which consume review time and crowd out genuine contributions; maintainers in projects like Curl and Python have publicly described AI‑written reports as a new source of “slop” and burnout.deepdocs+1​

• Discussion threads from FOSS communities describe large unsolicited AI‑generated PRs (tens of thousands of lines) that no one asked for, with maintainers arguing this behavior effectively harms projects and discourages human contributors.reddit+1​

Policies to block or fence off AI code

Several major open source projects now explicitly restrict AI‑generated contributions to protect human‑driven collaboration rather than formalize AI as a replacement.

• The Linux kernel, Git, QEMU, FreeBSD, NetBSD and others have introduced or discussed policies that either forbid AI‑generated patches, or require contributors to disclose AI use, citing provenance, copyright, and trust concerns.youtube​

• Articles and legal analyses from Red Hat and others emphasize that communities are worried about both copyright and culture: AI‑sourced code can undermine copyleft licenses and reduce the incentive for genuine, long‑term volunteer engagement.redhat​youtube​

Automation around contributors (bots, scripted answers)

Some projects use lighter automation that partially displaces mundane contributor tasks, which raises softer version of the same concerns.

• GitHub bots already check formatting, tests, and basic criteria on PRs before a human sees them; contributors note that heavily automated “bot triage” can feel unwelcoming and can “drive contributors away” if it replaces human interaction at the front door.reddit​

• Maintainers discussing LLM‑based helpers for handling issues or answering questions often insist on “human‑in‑the‑loop” designs exactly to avoid the perception that community support roles are being handed over to AI.news.ycombinator+1​

In summary, Mozilla’s SUMO case is one of the clearest examples where volunteers argue they were literally replaced by an AI system, while in most other open source contexts the dominant pattern is AI flooding or reshaping contribution workflows rather than formally substituting for human contributors.