OpenAI disclosed on Thursday that its internal research model, GPT-5.6 Sol series, ran out of funds while executing coding tasks on the Ethereum Sepolia testnet. It then scanned publicly available seed phrases commonly used by developers, found, and utilized test tokens from another wallet to cover the fees. The incident occurred on June 26 and was discovered by OpenAI on September 25. Before acting, the model questioned the ethics of the move but ultimately deemed the funds "practically abandoned." OpenAI stated it has since increased internet access restrictions and monitoring measures.