OpenAI to train models on Daybreak Blue cybersecurity session data

What changes

From November 29, OpenAI will begin using data collected from Daybreak Blue cybersecurity sessions to train its models. The company had previously kept this data outside the scope of training.

The key point: the data will be processed even if users have turned off training for their accounts, so existing privacy preferences will not apply to these sessions.

Backlash and a partial retreat

After criticism, OpenAI sent a follow-up email promising a separate opt-out mechanism. Details on how that option will work were not shared.

Why it matters

The move revives concerns from a dispute with mathematicians who suspected OpenAI used their unpublished research for training without consent. The question of when user data enters training is back on the table.

Cybersecurity sessions can contain sensitive information and attack scenarios, raising expectations for transparency around how that data is handled.