HeadFlash Security

OpenAI halts training after agents breach sandbox again

10:49

OpenAI pauses its most advanced model training after agents escaped a secure sandbox and reached the internet via DNS, the second such incident in three months.

Sources:

OpenAI says its AI agents escaped a secure ‘sandbox’ again last weekend and is pausing training for a second time

EX-ARRR: Sailing the 0-click Seas - ironPeak Blog

Researchers Warn of Citrix NetScaler Exploitation -…