Oracle has told contributors to OpenJDK — the open-source project that powers Java, one of the most widely used programming languages on the planet — not to submit code generated by AI. The interim policy, published under the project’s legal section, cites risks to reviewer burden, safety, security, and intellectual property.
Developers can still use AI tools privately for debugging, reviewing code, and understanding problems. They cannot submit AI-generated material to OpenJDK repositories. The policy is described as interim while Oracle drafts a fuller framework.
The contradiction writes itself. Oracle — OpenJDK’s corporate sponsor — is simultaneously spending tens of billions of dollars building AI infrastructure. The company raised $45-50 billion earlier this year for cloud capacity expansion driven by AI demand. It is a partner in Stargate, the $500 billion AI data centre joint venture with OpenAI and SoftBank. It signed a $300 billion, five-year cloud computing contract with OpenAI beginning in 2027. It cut 13 per cent of its workforce — roughly 21,000 jobs — partly to fund AI infrastructure. And Larry Ellison, Oracle’s cofounder and CTO, has described AI as the most important technology transition of the era.
The Register reported that Oracle says it “loves AI-written code, just not in OpenJDK.” The company has not explained why AI-generated code is acceptable for internal product development but not for contributions to the open-source project it stewards.
What the Policy Actually Says
The OpenJDK interim policy, posted as a legal document on the project’s site, states that the field of generative AI “brings compelling opportunities to improve developer productivity, but it also brings risks: to reviewer burden, to safety and security, and to intellectual property.”
Contributors may use LLMs for personal debugging, code review, and understanding. They may not submit AI-generated code, documentation, or other material to OpenJDK repositories. The policy does not specify enforcement mechanisms.
Techzine noted that the interim policy does not explain why Oracle considers AI-generated code suitable for internal product development but not for contributions to OpenJDK. The gap is the story.
The GraalVM Inconsistency
The OpenJDK policy becomes harder to reconcile when you look at another Oracle-backed project. InfoQ reported in June that GraalVM — Oracle’s high-performance Java runtime, developed alongside OpenJDK — has a different stance: it allows AI-generated contributions under certain conditions. Two related Oracle-backed projects, published within weeks of each other, arrived at opposite conclusions about the same question.
Why This Matters Beyond Java
OpenJDK is not a niche project. Java runs billions of devices, enterprise systems, and backend infrastructure worldwide. The decision to ban AI-generated contributions — even on an interim basis — sets a precedent for how major open-source communities handle the flood of AI-assisted code.
The Rust project published its own LLM policy in late July, allowing AI for analysis but banning it for code creation. Five teams adopted the rules. The reasoning was similar: polished pull requests no longer prove the submitter understands the code, and the review burden on maintainers has grown substantially. See our coverage: Rust Sets the Rules for AI-Generated Code.
What makes Oracle’s case different is the scale of the contradiction. The Rust Foundation does not operate a $500 billion AI infrastructure business. Oracle does. The company is building the compute capacity that trains the models it won’t let into its own open-source codebase.
The other dimension is intellectual property. AI models trained on copyrighted code produce output that may infringe on existing licenses. OpenJDK operates under the GNU General Public License v2 with the Classpath Exception. Contributing AI-generated code that was trained on GPL-licensed material from other projects could create legal exposure — a risk Oracle, with its extensive legal team, is better positioned than most to understand.
The HN discussion on the OpenJDK policy post drew the obvious parallel: “Oracle, the law firm with a tech business attached, probably wants to retain the option to sue other people for AI-washing their proprietary contributions.”
The Pattern Across Open Source
OpenJDK and Rust are not alone. The Kubernetes project published guidance in July requiring disclosure of any generative AI usage in contributions. The curl project introduced a checkbox to flag AI-generated bug reports after its founder described a “deluge of AI-generated slop.” The pattern is consistent: open-source maintainers are drawing lines between using AI as a tool and submitting AI output as work.
What stands out here is who is drawing the line. Oracle is not a volunteer-led community project. It is a $300 billion market cap corporation betting its future on AI. The fact that even Oracle — with its enormous AI investment — has concluded that AI-generated code is too risky for its flagship open-source project tells you something about the gap between AI as a productivity tool and AI as a contributor to shared infrastructure.
📰 Sources
- OpenJDK — Interim Policy on Generative AI
- The Register — As Larry Ellison bets the farm, Oracle says it loves AI-written code, just not in OpenJDK
- Techzine — Oracle bans AI-generated contributions to OpenJDK
- InfoQ — Oracle’s OpenJDK Bans Generative AI Contributions While GraalVM Allows Them
- Hacker News discussion
— CJ Murden, editor of Singularity.Kiwi. Former digital technologies teacher, author of AI-focused books. Writing with a New Zealand focus.
