Skip to content
K = 0.731Subscribe

OpenAI Plans Training on Daybreak Blue Sessions, Promises Separate Opt-Out After Backlash

1 reports 1 reporting sources Updated 2 hours ago

Understand the event

AI overview

OpenAI told Trusted Access members it plans to begin using previously excluded Daybreak Blue session data for model training on November 29, then promised a separate opt-out after users pushed back. The change would bring those sessions into its training data, while interactions flagged during safety reviews may still be used for safety purposes. Researchers raised concerns about unpublished work, citing the Buckmaster-Alpöge case. OpenAI denied training on the drafts submitted in that case.

Generated by AI from reports · Updated 49 minutes ago

Latest development11/1 00:38 (UTC+8)
OpenAI plans to train models on previously excluded Daybreak Blue cybersecurity session dataThe Decoder

Read next

More changes in today’s briefingConfigure the ChatGPT desktop app with Amazon Bedrock in AWS GovCloud (US)

Report timeline

Follow the reports to explore different sides of the event.

11/1
  1. The Decoder
    OpenAI plans to train models on previously excluded Daybreak Blue cybersecurity session data

    OpenAI told Trusted Access members it would begin using Daybreak Blue session data for model training on November 29, changing its previous exclusion policy. After backlash, a second email promised a separate opt-out, though interactions flagged during safety reviews may still be used for safety purposes. Researchers raised concerns about unpublished work, citing the Buckmaster-Alpöge case, in which OpenAI denied training on their submitted drafts.

Keep up with AI, one email a day

Model releases, price changes and compute developments, with original sources. Free, confirmed by email, and easy to unsubscribe. Read a past issue

Event heat trend

There is not enough continuous observation data to plot a trend yet.