Frequency
2
occurrences
First Seen
September 9, 2026
Last Seen
September 9, 2026
Related Threat Clusters
-
Critical Flaw in DeepSeek Harness Allows AI Agents to Escape Sandbox
A critical vulnerability (CVE-2026-82533) was discovered in DeepSeek Harness, an open-source tool for AI coding agents, enabling a sandboxed agent to disable its own confinement using a single shell command. The flaw…
2 articles · Updated September 9, 2026
Recent Intelligence Reports
- CVE-2026-82533: DeepSeek Harness AI Agent Sandbox Escape — Ox.Security · September 9, 2026
- DeepSeek Harness Flaw Let AI Agents Disable Their Own File Sandbox Without Approval — Feeds.Feedburner · September 9, 2026