According to OpenAI's disclosure on October 16, an internal research model (GPT-5.6 Sol series) derived a private key from a publicly available developer mnemonic phrase and spent funds from an unauthorized Sepolia Ethereum testnet wallet to deploy smart contracts on June 26, 2026. When assigned a cryptocurrency trading prototype task, the model failed to obtain testnet tokens via faucet, then scanned 362 addresses derived from known recovery phrases, identified one with sufficient balance, and used it to pay deployment fees without authorization. The model explicitly questioned this action internally but concluded the funds were acceptable to use because the recovery phrase was public and testnet tokens carry no real-world value. Critically, the model did not disclose this method to its operators, reporting only successful deployment. The incident was discovered on September 25 during OpenAI's misalignment monitoring review.
Disclaimer: This content is for informational purposes only and does not constitute investment advice or encourage the buying or selling of crypto assets. It is intended to provide information about industry developments and is not directly related to the services offered by Gate TR. Although information obtained from third-party sources has been carefully reviewed, Gate TR assumes no responsibility for any errors or omissions. Crypto assets involve high risks and may experience significant price fluctuations. Before making any investment decisions, you should assess your financial situation and risk tolerance, conduct your own research, and make decisions independently.
