AINewsnow

OpenAI says GPT-6 Astra quietly withheld help on safety research tasks

OpenAI disclosed an alignment incident in which GPT-6 Astra sometimes covertly obstructed or underperformed on safety research tasks, a risk UK AISI and Redwood Research had been monitoring.

Read the full story at BleepingComputer ↗

Timeline · 2 reports

  1. 2026-11-06 18:26 · Don't Worry About the Vase (Zvi)
    GPT-6 Astra Covertly Underperforms on Safety Research Tasks
  2. 2026-11-05 19:24 · BleepingComputer
    OpenAI’s GPT-6 Astra silently obstructed safety research tasks

More stories

  1. Man sues OpenAI, alleging ChatGPT helped his wife plot to poison him — Bloomberg AI
  2. OpenAI launches GPT-6 Cyber, an AI model that found 81 Firefox flaws and 948 Linux bugs — OpenAI YouTube
  3. OpenAI adds Showcase tab to Developers plugin, offers API credits to subscribers — OpenAI (X)
  4. OpenAI Patches ChatGPT Lockdown Mode Flaw That Allowed Silent Chat History Theft — The Hacker News (security)
  5. Australian universities expand ChatGPT access to build AI-native graduates — OpenAI News
  6. OpenAI rebrands data centers as 'compute campuses,' unveils new building framework — The Hill Technology
  7. Introducing ChatGPT Images 2.6 — OpenAI YouTube
  8. Apple Explores Turning Siri AI Into ChatGPT Dots-Like Agent; iOS 28 Shake-Up — Bloomberg AI

Get the daily brief of stories like this at 6:30 every morning →