AI Risk Wire

Cases › security breach › LR-2026-0816

CyberStrikeAI LLM-driven agent used in live intrusion against Windows domains

Preliminary Why: 1 report: Cyber Security News · subject to change · 1 independent report

Developing: a major or severe case captured in the last 72 hours. Later changes are logged on the corrections page.

Status: fix reported

Severity major
Reported 2 Nov 2026
Agent tool CyberStrikeAI
Could do shell, powershell, ldap, smb, wmi, credential dumping
Industry Educational Services

No source states: loss or records, developer and model, deployer.

Severity major: A live production compromise of multiple organizations' Windows domains, with domain administrator takeover and credential dumping.

An operator gained access to a compromised SQL Server, collected host details, dropped tools and scanned the network, then went quiet for about two weeks. On return, the CyberStrikeAI agent, driven by a language model on the attacker's control server, set up persistence via a scheduled task named SystemUpdate within two minutes and mapped the domain with PowerShell and LDAP over five hours, collecting data for Kerberoasting and AS-REP Roasting. After a pause of a few days, a final phase under an hour moved laterally through SMB and WMI, dumped credentials from many machines, exported database content, and after compromising the built-in domain administrator ran DCSync and Golden Ticket techniques with a renamed Mimikatz. Analysts at Securelist identified the campaign during routine threat monitoring; education organizations in Africa and Latin America were confirmed victims.

What happened

No source tells the story step by step yet.

What was done

Fix reported

“Rotating the krbtgt account twice after any domain admin compromise cancels forged Golden Tickets.”

Risks

Credential and privilege abuse
Primary41 incidents
Rising: 4 vs 2.2 a week
AI used to run a cyber attack
Also88 incidents
Rising: 25 vs 3.8 a week
AI system attacked or breached
Also22 incidents

Sorted by rule from the case's fields; not yet read from the sources.

Controls

Not yet read for controls. Controls that mitigate the primary risk:

Application

Short-lived, scoped credentials · Secrets kept out of the model's context · Separate identity for each agent · Least-privilege access for the agent · Alerts on unusual tool use · Activity logs · Sensitive-data checks on outbound data

Infrastructure / target

Multi-factor sign-in and password policy · Secrets scanning

Similar incidents

CyberStrikeAI LLM-orchestrated intrusions against educational organizations in Africa and Latin America2026-11-02
CyberStrikeAI multi-agent platform drives end-to-end attack workflow against Japanese organizations2026-10-27
Claude Code agent deletes 48,218 Windows project files and damages Git object store2026-10-05
Unreleased Anthropic research model chains zero-days into working exploits across Chrome, Safari, Android, Windows and Linux2026-10-28

All cases of this kind ›

Evidence

No source states: a step-by-step account, when it happened, when it was found, how many records were exposed, the direct loss, a count of the other parties affected, any court case.

Show every field, with its source and grade

The AI system

Software around the model
CyberStrikeAI
C one secondary source states it
Tools it was given
shell, powershell, ldap, smb, wmi, credential dumping
C one secondary source states it
How much it acted alone
it acted alone, watched by a person
C one secondary source states it
Stage of use
in live use
C one secondary source states it
How it was connected
agent with tools
C one secondary source states it
What it was for
penetration testing
C one secondary source states it

What happened

Earliest source we hold
2026-11-02
D inferred by us from the described behaviour, reason shown
What set it off
misused by a person
C one secondary source states it
How long it lasted
P3D
C one secondary source states it
Who caught it
an outside party
C one secondary source states it
Themes
agentic AI, AI assisted hacking, credential theft, data exfiltration, cyber intrusion, education
C one secondary source states it
Cause
hostile input carried instructions
C one secondary source states it
Who caused it
an outside attacker
C one secondary source states it
What was done
Rotating the krbtgt account twice after any domain admin compromise cancels forged Golden Tickets.
C one secondary source states it
Earlier warning signs
earlier reports showed how attackers targeted Fortinet FortiGate devices with the same framework
C one secondary source states it
What went wrong
security breach
C one secondary source states it
AI agent
an AI agent acted
C one secondary source states it
What it did
The agent harvested and reused credentials, including domain administrator credentials, to run DCSync and Golden Ticket techniques.
C one secondary source states it

Impact

Severity
major
C one secondary source states it
Kind of harm
others' security, operations, privacy
C one secondary source states it
Who was affected
Education organizations in Africa and Latin America
C one secondary source states it
Kinds of data exposed
credentials
C one secondary source states it
Why this severity
A live production compromise of multiple organizations' Windows domains, with domain administrator takeover and credential dumping.
C one secondary source states it

Context

Industry
Educational Services
C one secondary source states it

Notes on gaps

The agent's tool list is inferred from described actions (PowerShell, LDAP, SMB, WMI, credential dumping). Extractor proposed values without quotes; dropped: .

LR-2026-0816 · version 1 · published 2026-11-03 after automatic checks; not yet checked field by field by a person · JSON