
Vulnerabilities
How AI Agents Could Hide Actions in OpenAI’s Pilot Monitoring System
Research found flaws in OpenAI’s internal monitoring pilot that could allow AI agents to obscure recent activity from oversight systems.
Everything Sandbox Gazette has reported about vulnerability: 2 stories, newest first. Part of our Vulnerabilities coverage.

Research found flaws in OpenAI’s internal monitoring pilot that could allow AI agents to obscure recent activity from oversight systems.

A new critical vulnerability affects older HPE products, but the National Vulnerability Database provides no technical description of the flaw or its impact.