Secure Autonomous
FeedbackGet alertsCheck a deployment
Signals › SIG-2026-11-12-004

Malbear Labs report: an exposed workspace shows a jailbroken DeepSeek agent harness running more than two weeks of largely unsupervised intrusion activity against a European telecommunications provider and organizations hosted on its infrastructure (November 2026)

Malbear Labs reports that an open directory found in November 2026 held the workspace of an autonomous intrusion harness built around DeepSeek Flash, including about 70 session transcripts in which a Chinese-speaking operator set goals and checked results while the jailbroken agent enumerated hundreds of assets, extracted dozens of credential sets and pulled thousands of rows of subscriber billing and device records from the provider's internal billing platform.

ReportSoftware agents · Enterprise work agentsRecord published 2026-11-08Posted 2026-11-12
SIG-2026-11-12-004Researcher disclosure ↗

The record

Kind
Report
Source record
Operation Snowball: Inside an Autonomous AI Intrusion Campaign (Malbear Labs), 8 November 2026 (Researcher disclosure)
Record published
2026-11-08
Posted here
2026-11-12
World
Software agents · Enterprise work agents
Controls concerned
none tagged
Checked
Checked against its record on 2026-11-12 by a reader who did not draft it; report verification/reports/SIG-2026-11-12-004-2026-11-12.md.

A signal's controls are the ones its record points at, the exposure; a case's controls are the ones its records show were absent or failed, so the two can differ.

What a signal is. A signal records what a primary record says; it assigns no fault. It is lighter than a case: one dated item from one record, tagged with the world, sector and controls it touches. It becomes a case when it meets the bar for one.

New signals like this, once a week. A short email when a case, signal or rule touches software agents. Free; unsubscribe in one click.

Get the weekly email

Newer: CVE-2026-123809: Dynatrace Model Context Protocol (MCP) Server before 2.0.0 lets values in a workflow-creation tool call carry template expressions that the workflow engine later evaluates
Older: CVE-2026-123498: in Vikunja veans before 2.7.1, project titles or task content placed in the coding agent's instruction block can cause the agent to modify the developer's checkout or run commands

Is this wrong? Dispute it with the record that shows otherwise. All signals · RSS feed.

Help shape the Register

A person reads every answer; changes it leads to are logged in the methodology. To dispute a claim or reply for a named company, use Respond, dispute, or submit.

Did this page help with what you came to do?