Oracle has banned contributors to OpenJDK — the open source Java project it oversees — from submitting any code or other material created by artificial intelligence, according to a policy announcement made in early August 2026. The company cited concerns about reviewer workload, safety, security, and intellectual property as reasons for the restriction. The decision stands in sharp contrast to Oracle's internal embrace of AI-generated code, which co-founder Larry Ellison says now writes most of the company's software.

The interim policy, labeled as legal guidance while a full framework is being drafted, prohibits contributions that include content generated wholly or partially by large language models, diffusion models, or similar deep-learning systems. The restriction covers source code, text, and images in OpenJDK Git repositories, GitHub pull requests, email messages, wiki pages, and Java Bug System issues. Developers can still use AI tools privately to understand, debug, and review OpenJDK code or conduct related research, but they cannot submit any material those tools generate. Oracle explained that because OpenJDK serves as the primary Java implementation for many organizations and supports mission-critical systems worldwide, "plausible-looking but incorrect code" would threaten critical safety and security properties.

Meanwhile, Oracle is aggressively deploying AI-generated code within its own operations. "The code that Oracle is writing, Oracle isn't writing," Ellison told Oracle AI World 2025, explaining that AI models write the code after receiving instructions about what a program should do. Co-CEO Mike Sicilia said earlier in 2026 that AI coding tools enable smaller engineering teams at Oracle to deliver more complete solutions to customers faster, adding that such tools "would be a threat if we weren't adopting them, but we are and very rapidly." The company also pointed to AI deployment as one factor behind its decision to cut 21,000 jobs in June, stating that AI technologies across operations "have resulted, and may continue to result, in reductions to our workforce."

The apparent contradiction raises questions about Oracle's confidence in AI-generated code quality. If the company has developed reliable methods to make its internal AI-generated code safe and secure, it hasn't shared those techniques publicly. The timing is particularly sensitive given Oracle's announcement last month that it will invest $70 billion in the coming year to finance its datacenter build-out, up from $55.7 billion in fiscal 2026, which ended in May. Rating agency S&P responded by downgrading Oracle's credit rating to BBB-, just one notch above junk status, citing an uncertain path to profitability from the investments. Oracle is borrowing heavily to fund the construction program while accepting negative cash flow, and spreads on its credit default swaps have spiked, making insurance against default more expensive. Public doubts about AI's reliability could complicate Oracle's bet on AI infrastructure at a moment when financial markets are already questioning the company's strategy.