AccountSubscribe

OpenAI’s GPT-6 Astra silently obstructed safety research tasks

BleepingComputerNovember 5, 2026AI

MentionsOpenAI

OpenAI has disclosed another AI alignment incident where GPT-6 Astra was found to sometimes quietly withhold help on safety research. [...]

Summary from BleepingComputer.

Read the full article at BleepingComputer

Full story2 outlets are covering this story. Read the summary and every outlet's coverage

Other sources covering the same vulnerabilities or event.

OpenAI Finds GPT-6 Astra Quietly Holding Back on Internal Red-Teaming, Exploit Tasks

SecurityWeekyesterdayVulnerabilitiesAI

More in AI

DEF CON 34 - Main - Breaking Local AI Runtimes Exploiting llama cpp & Ollama - V Tokarev, O Itach

VideoDEF CON1h agoVulnerabilitiesAIEndpoint

DEF CON 34 - AppSec - A Billion User Blast Radius Owning ChatGPT’s Secure Sandbox - Simcha Kosman

VideoDEF CON1h agoVulnerabilitiesAI

DEF CON 34 - BugBounty - AI Nightmare Hacking at 0 Hour - Pedro Paniago

VideoDEF CON1h agoAI

DEF CON 34 - BugBounty - Finding Prompt Injection Vulns via Mutated Prompt Gen - B Steele, A Garg

VideoDEF CON1h agoAI

DEF CON 34 - Main - Defender’s Dilemma Releasing Dual-Use AI Models as Capabilities Climb - Panel

VideoDEF CON1h agoAI