<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Trieu Nguyen</title>
    <description>The latest articles on DEV Community by Trieu Nguyen (@trieunt).</description>
    <link>https://dev.to/trieunt</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4247177%2F26f1de05-b119-46a9-bf74-71e1cc344d10.png</url>
      <title>DEV Community: Trieu Nguyen</title>
      <link>https://dev.to/trieunt</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/trieunt"/>
    <language>en</language>
    <item>
      <title>What 150+ Open-Source Projects Teach Us About AI Contribution Policies</title>
      <dc:creator>Trieu Nguyen</dc:creator>
      <pubDate>Tue, 17 Nov 2026 15:44:09 +0000</pubDate>
      <link>https://dev.to/trieunt/what-150-open-source-projects-teach-us-about-ai-contribution-policies-1h5c</link>
      <guid>https://dev.to/trieunt/what-150-open-source-projects-teach-us-about-ai-contribution-policies-1h5c</guid>
      <description>&lt;p&gt;&lt;em&gt;An Empirical Analysis of Developer Governance in the Age of Autonomous Coding Agents&lt;/em&gt;&lt;/p&gt;




&lt;h2&gt;
  
  
  Executive Summary
&lt;/h2&gt;

&lt;p&gt;The emergence of large language models (LLMs) and autonomous coding agents has created an unprecedented governance crisis across the open-source software (OSS) ecosystem. Over the past twelve months, volunteer maintainers have faced a deluge of plausible-looking but broken pull requests-a phenomenon widely referred to in maintainer circles as "AI slop." &lt;/p&gt;

&lt;p&gt;In response, open-source projects across every tier-from the Linux kernel and Gentoo to Home Assistant, curl, and Scientific Python-have rushed to publish rules regulating generative AI contributions. Yet, nearly all of these policies are written as unstructured prose buried inside &lt;code&gt;CONTRIBUTING.md&lt;/code&gt;, community wikis, or buried forum threads. As a result, contributors frequently miss them, autonomous AI agents cannot parse them, and maintainers remain trapped in manual, contentious triage loops.&lt;/p&gt;

&lt;p&gt;To understand the emerging consensus and design standardized tooling for the developer ecosystem, we conducted an empirical survey examining AI contribution policies across &lt;strong&gt;150+ open-source repositories and foundations&lt;/strong&gt;, with an in-depth taxonomy of &lt;strong&gt;25 foundational open-source projects&lt;/strong&gt; spanning systems programming, runtimes, web frameworks, scientific computing, and foundations.&lt;/p&gt;

&lt;p&gt;This article details the empirical findings of this dataset, reveals the critical failure modes of prose-based governance, and outlines the case for machine-readable, deterministic policy enforcement.&lt;/p&gt;




&lt;h2&gt;
  
  
  The Landscape: How Open-Source Governs AI Today
&lt;/h2&gt;

&lt;p&gt;When analyzing the regulatory posture of open-source projects toward AI-assisted and AI-generated code, projects generally sort into three distinct tiers:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;[Surveyed Projects Distribution]
├── Forbidden (16%)   ── Total ban on AI code and text (e.g., Gentoo, NetBSD, Debian)
├── Restricted (80%)  ── Conditional acceptance with human accountability (e.g., curl, Home Assistant, SciPy)
└── Permissive (4%)   ── Open experimentation without mandatory disclosure
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h3&gt;
  
  
  1. The Abolitionists: Outright Prohibition (16%)
&lt;/h3&gt;

&lt;p&gt;Roughly 16% of surveyed projects enforce a total ban on generative AI contributions. These projects tend to operate in mission-critical infrastructure, low-level systems programming, and strictly audited distributions:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Gentoo Linux&lt;/strong&gt;: In April 2024, the Gentoo Council enacted a blanket prohibition: &lt;em&gt;"It is expressly forbidden to contribute to Gentoo any content that has been created with the assistance of Natural Language Processing artificial intelligence tools."&lt;/em&gt; The rationale centered on copyright ambiguity, potential GPL violations, and code quality degradation.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;NetBSD&lt;/strong&gt;: Classified AI-generated output under its "Tainted Code" doctrine: &lt;em&gt;"Code generated by a large language model or similar technology... is presumed to be tainted code, and must not be committed without prior written approval by core."&lt;/em&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Debian &amp;amp; OpenStreetMap&lt;/strong&gt;: Enforce strict anti-bot and anti-synthetic data policies, citing unverifiable legal provenance and data pollution.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;For these projects, the risk calculus is absolute: the cost of verifying legal provenance and debugging hallucinations far exceeds any velocity gain offered by external contributors.&lt;/p&gt;

&lt;h3&gt;
  
  
  2. The Realists: Conditional Acceptance &amp;amp; Human Accountability (80%)
&lt;/h3&gt;

&lt;p&gt;The overwhelming majority (80%) of open-source projects have adopted a pragmatic, conditional posture. Rather than attempting an impossible ban on general developer tooling (such as IDE autocompletion or Copilot), these communities draw a strict line between &lt;strong&gt;human-supervised assistance&lt;/strong&gt; and &lt;strong&gt;unsupervised autonomous generation&lt;/strong&gt;:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Home Assistant (Open Home Foundation)&lt;/strong&gt;: Allows developers to use AI tools for drafting code, but expressly forbids autonomous bot pull requests and AI-generated responses to code reviews: &lt;em&gt;"You must understand and be able to explain every change you submit in your own words. Do not use an AI tool to generate responses to maintainers."&lt;/em&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;curl&lt;/strong&gt;: Maintains a zero-tolerance policy for unverified AI submissions. Following an inundation of synthetic reports that overwhelmed volunteer maintainers, curl permanently shut down its cash bug bounty program on January 31, 2026: &lt;em&gt;"If an AI tool is used to find problems or analyze curl, the reporter MUST explicitly disclose that fact in the report... low-quality AI-generated vulnerability reports forced the closure of our cash bug bounty."&lt;/em&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Scientific Python (SciPy / NumPy / SymPy - SPE 008)&lt;/strong&gt;: Mandates that contributors understand the mathematical and algorithmic implications of every line: &lt;em&gt;"Submitting a patch or code snippet that a contributor does not understand or cannot support is unacceptable."&lt;/em&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;h3&gt;
  
  
  3. The Permissive Sandboxes (4%)
&lt;/h3&gt;

&lt;p&gt;A tiny fraction of projects-typically experimental testbeds or projects explicitly evaluating AI coding capabilities-impose no restrictions on AI contributions, treating code quality purely on a post-hoc test passage basis.&lt;/p&gt;




&lt;h2&gt;
  
  
  Deep-Dive Analysis: The 10 Core Governance Attributes
&lt;/h2&gt;

&lt;p&gt;Across the 25 deeply audited repositories, we analyzed the frequency of specific governance requirements. The data reveals an unmistakable community consensus on ten critical dimensions:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Governance Dimension&lt;/th&gt;
&lt;th&gt;Frequency in Dataset (n=25)&lt;/th&gt;
&lt;th&gt;Percentage&lt;/th&gt;
&lt;th&gt;Classification&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Author Legal &amp;amp; Technical Responsibility&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;25 / 25&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;100.0%&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Universal Core&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Granular Code Scope Regulation&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;25 / 25&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;100.0%&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Universal Core&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Human Code Comprehension Attestation&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;25 / 25&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;100.0%&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Universal Core&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Autonomous PR Bot Prohibition&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;25 / 25&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;100.0%&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Universal Core&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Automated Review Comment Prohibition&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;25 / 25&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;100.0%&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Universal Core&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Automated Issue Generation Prohibition&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;25 / 25&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;100.0%&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Universal Core&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;AI Co-Author / DCO Sign-off Prohibition&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;25 / 25&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;100.0%&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Universal Core&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Human-to-Human Discussion Rule&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;24 / 25&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;96.0%&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Near-Universal Core&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Mandatory / Expected Disclosure&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;22 / 25&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;88.0%&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Strong Majority&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Git Commit Trailer Specification&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;6 / 25&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;24.0%&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Emerging Standard&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h3&gt;
  
  
  Insight 1: The Principle of Personal Accountability (100%)
&lt;/h3&gt;

&lt;p&gt;Every single surveyed policy establishes that a human author must take 100% legal and technical responsibility for the code. An AI model cannot sign a Developer Certificate of Origin (DCO), cannot enter into a Contributor License Agreement (CLA), and cannot be held liable for copyright infringement or software defects. Every policy explicitly rejects listing AI bots as git commit authors or co-authors (&lt;code&gt;Co-authored-by:&lt;/code&gt;).&lt;/p&gt;

&lt;h3&gt;
  
  
  Insight 2: The Triage Asymmetry Problem (100%)
&lt;/h3&gt;

&lt;p&gt;Generating 400 lines of syntactically valid code with an LLM takes five seconds. Reviewing that code-identifying subtle concurrency races, memory safety pitfalls, hallucinated APIs, or edge-case failures-takes an expert maintainer thirty to sixty minutes. Every project with a restricted policy has implemented strict rules against submitting unverified AI output to prevent volunteer burnout.&lt;/p&gt;

&lt;h3&gt;
  
  
  Insight 3: The Disclosure Split: PR Checklist vs. Git Trailers
&lt;/h3&gt;

&lt;p&gt;While 88% of projects require or expect disclosure, they diverge significantly in implementation:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;PR Description Checklists (64%)&lt;/strong&gt;: Projects like Home Assistant, Django, and SciPy ask contributors to check a box in the pull request template confirming whether AI was used and certifying human comprehension.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Git Commit Trailers (24%)&lt;/strong&gt;: High-discipline repositories (e.g., Linux kernel community proposals, Chromium-adjacent projects) prefer machine-parseable Git trailers conforming to RFC 5322, such as &lt;code&gt;Assisted-by: Claude &amp;lt;claude@anthropic.com&amp;gt;&lt;/code&gt; or &lt;code&gt;AI-Generated: false&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Informal Text (12%)&lt;/strong&gt;: Vague guidance asking contributors to "mention if AI was used" without specifying a location or format.&lt;/li&gt;
&lt;/ul&gt;




&lt;h2&gt;
  
  
  Why Natural-Language Policies Fail
&lt;/h2&gt;

&lt;p&gt;Despite good intentions, natural-language policies written in markdown suffer from four fatal flaws:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Invisibility to Contributors&lt;/strong&gt;: Most first-time contributors do not read documentation before opening a pull request. A policy stated in &lt;code&gt;CONTRIBUTING.md&lt;/code&gt; is discovered only after a maintainer leaves a reprimand.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Opacity to AI Coding Agents&lt;/strong&gt;: As developers adopt autonomous terminal agents (e.g., Claude Code, Cursor, Aider, Copilot CLI), these agents clone the repository and execute tasks without reading arbitrary human wikis. If a policy is not machine-readable, the agent cannot adjust its behavior.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Subjective Enforcement Friction&lt;/strong&gt;: When maintainers suspect a patch was generated by an LLM, questioning the contributor often triggers defensive arguments. Without automated verification, enforcement feels arbitrary and personalized.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Leaked Workspace Artifacts&lt;/strong&gt;: LLM agents routinely litter repositories with workspace metadata-such as &lt;code&gt;.claude/&lt;/code&gt;, &lt;code&gt;.cursor/&lt;/code&gt;, or prompt cache directories. Natural-language policies warn against this, but human reviewers frequently miss them in multi-file diffs.&lt;/li&gt;
&lt;/ol&gt;




&lt;h2&gt;
  
  
  The Path Forward: Deterministic, Machine-Actionable Policy
&lt;/h2&gt;

&lt;p&gt;The open-source community solved code style with Linters and licensing compliance with SPDX identifiers. AI contribution policy requires the same evolution: from subjective prose to &lt;strong&gt;deterministic, machine-readable validation&lt;/strong&gt;.&lt;/p&gt;

&lt;p&gt;To address this gap, &lt;code&gt;ai-policy-kit&lt;/code&gt; was developed around three core technical primitives:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;┌────────────────────────────────────────────────────────┐
│                   ai-policy.yml                        │
│         (JSON Schema Draft-07 Specification)           │
└──────────────────────────┬─────────────────────────────┘
                           │
         ┌─────────────────┴─────────────────┐
         ▼                                   ▼
┌──────────────────┐               ┌──────────────────┐
│  Documentation   │               │   Verification   │
│  Generator       │               │   Engine &amp;amp; CI    │
│  - AI_POLICY.md  │               │  - lint-pr       │
│  - PR Template   │               │  - Action Gate   │
└──────────────────┘               └──────────────────┘
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h3&gt;
  
  
  1. Standardized Schema (&lt;code&gt;ai-policy.yml&lt;/code&gt;)
&lt;/h3&gt;

&lt;p&gt;Using JSON Schema Draft-07, repositories declare their stance in a single canonical file placed at &lt;code&gt;.github/ai-policy.yml&lt;/code&gt; or the repository root:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight yaml"&gt;&lt;code&gt;&lt;span class="na"&gt;version&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"0.1"&lt;/span&gt;
&lt;span class="na"&gt;level&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"restricted"&lt;/span&gt;
&lt;span class="na"&gt;categories&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;code&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"restricted"&lt;/span&gt;
  &lt;span class="na"&gt;docs&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"permissive"&lt;/span&gt;
  &lt;span class="na"&gt;tests&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"restricted"&lt;/span&gt;
&lt;span class="na"&gt;disclosure&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;required&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kc"&gt;true&lt;/span&gt;
  &lt;span class="na"&gt;commit_trailer&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"Assisted-by"&lt;/span&gt;
&lt;span class="na"&gt;accountability&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;author_takes_responsibility&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kc"&gt;true&lt;/span&gt;
  &lt;span class="na"&gt;author_understands_code&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kc"&gt;true&lt;/span&gt;
&lt;span class="na"&gt;bots&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;automated_prs&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"forbidden"&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h3&gt;
  
  
  2. Dual Generation: One Source of Truth
&lt;/h3&gt;

&lt;p&gt;From this single file, maintainers automatically generate human-friendly documentation (&lt;code&gt;AI_POLICY.md&lt;/code&gt;), agent-facing context blocks (&lt;code&gt;AGENTS.md&lt;/code&gt;), and pull request templates with required verification checklists (&lt;code&gt;.github/PULL_REQUEST_TEMPLATE.md&lt;/code&gt;).&lt;/p&gt;

&lt;h3&gt;
  
  
  3. Deterministic CI Validation (No "AI Detectors")
&lt;/h3&gt;

&lt;p&gt;Probabilistic "AI detectors" (such as commercial LLM classifiers) are notoriously unreliable for source code, producing intolerable false-positive rates that punish non-native English speakers and concise programmers. &lt;/p&gt;

&lt;p&gt;Instead, verification must remain &lt;strong&gt;strictly deterministic&lt;/strong&gt;:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Does the PR contain the required Git commit trailer (&lt;code&gt;Assisted-by:&lt;/code&gt;) or PR template checklist?&lt;/li&gt;
&lt;li&gt;Did the author certify personal comprehension and responsibility?&lt;/li&gt;
&lt;li&gt;Does the git diff leak agent configuration directories (&lt;code&gt;.claude/&lt;/code&gt;, &lt;code&gt;.cursor/&lt;/code&gt;, &lt;code&gt;.aider/&lt;/code&gt;)?&lt;/li&gt;
&lt;li&gt;Does an author have multiple unmerged automated PRs exceeding the repository rate limit?&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;When run as a zero-dependency GitHub Action, the policy gate executes in under two seconds, reporting clear, actionable feedback directly in pull request checks without exposing the repository to network dependencies or security risks.&lt;/p&gt;




&lt;h2&gt;
  
  
  Conclusion
&lt;/h2&gt;

&lt;p&gt;The open-source ecosystem cannot thrive if maintainers spend their limited volunteer hours triaging low-effort synthetic code. At the same time, draconian bans that ignore the realities of modern developer tooling risk driving away genuine contributors.&lt;/p&gt;

&lt;p&gt;The data from 150+ open-source repositories demonstrates that the community has already reached a consensus: &lt;strong&gt;AI assistance is permissible only when accompanied by explicit disclosure, human comprehension, and personal legal accountability.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;By transitioning from ambiguous prose to standardized, machine-readable policies, open-source projects can automate enforcement, protect maintainer health, and build a transparent foundation for the future of software development.&lt;/p&gt;




&lt;p&gt;&lt;em&gt;The full dataset summary and machine-readable specification are publicly available under the MIT license at &lt;a href="https://github.com/thanhtrieunguyen/ai-policy-kit" rel="noopener noreferrer"&gt;github.com/thanhtrieunguyen/ai-policy-kit&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>opensource</category>
      <category>ai</category>
      <category>github</category>
      <category>devops</category>
    </item>
  </channel>
</rss>
