OpenAI to train models on previously excluded Daybreak Blue cybersecurity data

openai will start using daybreak blue session data to train its models on november 29, according to an email sent to trusted access program members. previously, this data was excluded from training, even when users had opted to 'improve the model for everyone'. the change has sparked concerns among researchers, who fear a repeat of the buckmaster-alpöge case, where openai was suspected of using sensitive information without consent. openai has since promised a separate opt-out for users, but interactions flagged during safety reviews can still be used for safety purposes.