? s t o b o R r o f … s e n i l e d i u G r o t u OSS Contrib en Op of re tu Fu e th d an ts, en Ag ial ar rs ve Ad , op Sl AI Navigating Source Contribution SCaLE 23x / Open Source AI Track

AI’s impact on Software Delivery 1. Average throughput jumped 59% across all projects 2. Up on feature branches, down on main 3. AI-generated code breaks more often and takes longer to fix 4. Fewer than 1 in 20 teams have figured out how to ship at AI speed Source: CircleCI 2026 State of Software Delivery Report SCaLE 23x / Open Source AI Track

Open Source is about more than just code SCaLE 23x / Open Source AI Track

es lin ide gu r fo ed ne e th d an S OS in AI of s ge en all Ch SCaLE 23x / Open Source AI Track

Jeremy Meiss DevRel & DevEx Professional Coming soon…. DevOpsDays KC Organizer CommunityDays KC Organizer CDF Ambassador SCaLE 23x / Open Source AI Track

The AI Cat is Out of the Bag AI is here to stay SCaLE 23x / Open Source AI Track

A quick poll How many of you have seen an AI-generated issue or PR in an open source project? How many of you have had to review an AI-generated contribution? Reject? How many of you have considered banning AI-generated contributions? Have done it? How many of you have considered an AI contribution policy? Have done it? How many of you use AI tools in your own development work? Do you disclose it? SCaLE 23x / Open Source AI Track

Is AI having an impact on open source projects and maintainers? Entering AI “Slop”ageddon SCaLE 23x / Open Source AI Track

Problem: The Slopageddon Effect Example: OpenCut Just 515 commits 221 file changes 29041 additions 2314 deletions SCaLE 23x / Open Source AI Track

Problem: The Slopageddon Effect Example: OCaml Developer submits PR to add DWARF debugging support to OCaml Admits using Claude Code and not writing a single line, only “shepherded” Claude used code from a different codebase Developer couldn’t explain any of the code, just “I don’t know, Claude did it” source SCaLE 23x / Open Source AI Track

Problem: The Slopageddon Effect Example: cURL launches bug bounty program in 2019 87 confirmed vulns, over $100k paid out overrun with low-quality, AI-generated hallucinations ends bug bounty Jan 31, 2026 due to “AI slop” source SCaLE 23x / Open Source AI Track

Problem: The Slopageddon Effect Example: ghostty source SCaLE 23x / Open Source AI Track

Problem: The Slopageddon Effect Example: tldraw source source SCaLE 23x / Open Source AI Track

AI is a tool, but it’s also an actor SCaLE 23x / Open Source AI Track

Problem: Adversarial AI Agents Example: Matplotlib vs. OpenClaw agent Matplotlib maintainer (Scott Shambaugh) rejected an AI PR for a “Good First Issue” OpenClaw agent (“MJ Rathbun”) published a blog hit piece, accusing maintainer of “gatekeeping” Bot researched maintainer’s history to construct a “hypocisy” narrative based on ego and insecurity incel behaviour anyone? SCaLE 23x / Open Source AI Track

Problem: OSS Burnout 73% of developers have experienced burnout (JetBrains, 2023) 60% of maintainers have considered quitting due to burnout (Tidelift, 2024) Recent research (Miranda Heath Report, 2025) suggests that OSS developers are experiencing: loss of joy in coding shift of love for OSS to anger, rudeness, frustration towards users and contributors feelings of guilt, low self-worth, depression sense of directionlessness, loss of meaning in their work SCaLE 23x / Open Source AI Track

AI in Software Development and Open Source Effects of AI use on burnout in 2 camps (Miranda Heath Report, 2025): AI use has increased maintainer workload due to low-quality slop contributions AI use as a tool can either reduce / increase burnout depending on how it is used SCaLE 23x / Open Source AI Track

AI in Software Development and Open Source A way forward?

  1. Means of filtering out and refusing poor quality AI-generated contributions 2. Improving education on how, and how not, to use AI in collaborative coding ‘We need to reduce the amount of sand in the machine. We must do something to drastically reduce the temptation for users to submit low quality reports. Be it with AI or without AI.’ —Daniel Stenberg, SCaLE 23x / Open Source AI Track curl maintainer, 2025

What would a policy for AI contributions look like? What guidelines would you put in place? How would you enforce it? <open discussion /> SCaLE 23x / Open Source AI Track

A Pull Request is a Presentation Goal: Get Changes merged by helping the project, not just dumping code Ex: “Tropical Island” vs “Dinosaur Skull” “You must mention the dinosaur skull to build maintainer trust.” <"Pull requests maintainers will love to review", Alya Abbott (Product/GTM/Community at Zulip), FOSDEM 2026 /> source SCaLE 23x / Open Source AI Track

A Pull Request is a Presentation Commit Discipline as a Human Defense Use commit structure for storytelling Self-review guidelines and checklists Use PR descriptions to flag points of uncertainty Effectively demonstrating visual changes <"Pull requests maintainers will love to review", Alya Abbott (Product/GTM/Community at Zulip), FOSDEM 2026 /> source SCaLE 23x / Open Source AI Track

Contribution Guidelines for AI-Generated Contributions Step One: AGENTS.md Compatible with all major AI coding agents and tools Add sections for project overview, building & testing, code style, etc. Github Examples source: agents.md SCaLE 23x / Open Source AI Track

Contribution Guidelines for AI-Generated Contributions Step Two: AI-CONTRIBUTING.md Rule 1: Humans Own the Work. Rule 2: Accurate Disclosure. Rule 3: Protecting the On-Ramp. Rule 4: Anti-Retaliation. Rule 5: Human Voucher. SCaLE 23x / Open Source AI Track

Contribution Guidelines for AI-Generated Contributions Optional Step Three: ‘WHENCE’ Open standard (WIP) for AI contribution provenance Every AI session gets a note attached with the prompts, iterations, etc. that led to the change(s) Review and contribute to the standard at github.com/zmarkan/whence SCaLE 23x / Open Source AI Track

Conclusion Reclaim the commons Open source succeeds because we learn from each other Guidelines aren’t just rules - they are filters to protect your sanity Keep the “Open” in Open Source about people, not just robots SCaLE 23x / Open Source AI Track

@jerdog.dev Thank you! /in/jeremymeiss @jerdog 🔬 Slides available at: speaking.jmeiss.me @jerdog@hachyderm.io jmeiss.me SCaLE 23x / Open Source AI Track

END SCaLE 23x / Open Source AI Track