Ledger / Oct 2026
Suspected China-nexus operator breaches 25 servers with possible AI assistance
Suspected
Dream found Claude Code, an aiscan deployment configured for DeepSeek and Kimi, and consistently formatted scripts. It explicitly reported no direct evidence that AI wrote the exploits or post-exploitation scripts.
- Model
- Claude Code, DeepSeek, KimiAnthropic · closed; DeepSeek · openness unknown; Moonshot AI · openness unknown
- Agent harness
- Claude Code, aiscancustom framework
- Attributed country
- Chinalanguage or infrastructure clues only
- Sectors hit
- government, enterprise broad25 firewalls and servers, including three systems in a US government-affiliated…
The operator chained PHP2Shell vulnerabilities with the Fragnesia Linux privilege-escalation flaw. AI tooling and script artifacts suggest assistance, but do not establish agent-driven exploitation.
Sources
- Coverage of Dream Research's PHP2Shell firewall intrusion investigationThe Injection · 26 Oct 2026
- Date
- Oct 2026
- Disclosed
- 26 Oct 2026
- Actor
- Unidentified suspected China-nexus operator (unknown), as attributed by The Injection
- Nexus
- Suspected China nexus based on Chinese-language prompts and infrastructure settings; state sponsorship unconfirmed
- Targets
- 25 firewalls and servers, including three systems in a US government-affiliated environment
- What the AI did
- Exploit dev, Ops automation
- Model as named
- Claude Code, DeepSeek, Kimi
- Model basis
- inferred
- Autonomy
- assist
- Outcome
- Attackers obtained root access to 25 hosts, established webshells and reached internal network and authentication information. Outcomes of further Active Directory attacks were unknown.