A later notice promises a separate opt-out for personal accounts, while API access for individuals still ends on November 20th.
By [Ryan Merket](https://runtimewire.com/author/ryan-merket)
· Published
Primary source: [OpenAI Help Center](https://help.openai.com/en/articles/7730893-data-controls-in-chatgpt)
Why it matters #
Daybreak Blue sessions can contain unpatched vulnerabilities, exploit proofs of concept, and incident data. Whether personal-account researchers can keep that material out of model improvement determines whether the program stays usable for independent security work, and the November 20th API cutoff pushes those who need programmatic access toward business accounts.
OpenAI told Daybreak Blue users on personal accounts that they will get a separate way to opt out of model improvement for Daybreak sessions, hours after first telling them that no such opt-out would apply from November 29th.
Interactions flagged for safety review may still be used for safety purposes, including work intended to stop the advanced cyber capabilities from being used for harm, according to the second notification sent to users. OpenAI said more detail would come closer to November 29th.…
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.