Oracle bans AI-generated code from OpenJDK
TL;DR Highlight
OpenJDK bans LLM-generated code contributions, sparking debate over AI policy in open source.
Who Should Read
Developers who submit AI-assisted PRs to open source or must set their team's AI code policy
Core Mechanics
- OpenJDK blocked LLM-generated code
- Community contributions must not include content generated in part or in full by large language models.
- The document is called the OpenJDK Interim Policy on Generative AI.
- The policy is interim
- The policy page says lawyers are writing the final version.
- Non-LLM editor features are still allowed
- Spell-check, grammar-check, auto-completion and refactoring are fine if not based on LLMs or deep learning.
- The original article was unavailable
- Only comments were provided, so details are limited to what the comments quote.
Community Reactions
- Opinion: it's about copyright
- Commenters said AI output may not be copyrightable, so copyright can't be assigned to Oracle.
- Another said big companies get indemnity from AI vendors, but external contributions lack that.
- Opinion: it's about review burden
- Commenters said low-care contributions add load to limited human reviewers.
- Another said code is a liability for a mature project.
- Oracle didn't issue the ban directly
- A commenter noted the OpenJDK Governing Board issued it, separate from Oracle's internal rules.
- Question about detection
- One commenter asked how incoming AI code could be detected and whether tools exist.
- Another noted Rust has similar guidelines.
How to Apply
- Draft your team's AI code policy
- Spend 30 minutes today writing one page on allowed scope, review standards and disclosure.
- Split rules for internal code and external contributions.
- Check AI policy before contributing upstream
- Read the target repo's CONTRIBUTING and AI policy before opening a PR.
- Don't include LLM-generated code in projects that ban it.
- Add an AI-usage checkbox to your PR template
- Add a field for whether and how much an LLM was used.
- Require the author to confirm they reviewed and understand every line.
Try It
# Paste into Claude or ChatGPT (simplified version for drafting a policy, not the article's method)
You are an open source maintainer. Write a one-page 'AI-generated code contribution policy' draft for this situation.
- Project: {{project description}}
- Number of reviewers: {{reviewers}}
- Concerns: copyright, review burden, code quality
Include: allowed scope, banned scope, exceptions for editor features (autocomplete, refactoring), and a checklist for PR authors.
Next: turn the finished draft into PR template checkboxes.Terminology
Related Papers
The Judge Can Debias Itself: Awareness and Self-Debiasing for Reliable LLM-as-a-Judge
Warning an LLM judge about bias or letting it self-critique improved 20 of 21 model–bias combinations.
90% of US working hours are spent on tasks not covered well by AI benchmarks
Mapping 29 benchmarks to job tasks shows 90% of US working hours lack good coverage
Postgres 19 delayed after AI reviewers find more bugs than humans can fix
LLM code review surfaced so many bugs that Postgres 19 slipped to November and 8-10 features were reverted
Show HN: Gemini 4 Argon plays StarCraft
LLMs wrote StarCraft bots in one hour; GPT-6 Astra scored highest.
A Plausible Explanation Is Enough to Mislead Multimodal Judges
Adding plausible explanations to fixed answers cut MLLM judges' wrong-answer rejection by 12.7–13.8 points.
Neural networks are clean
Looking at neurons directly shows most have one main job, challenging superposition.