Google's threat intelligence arm reported that Chinese and North Korean hackers deployed the AI-powered penetration testing tool Cairn in an intrusion targeting the U.S. aerospace sector. The findings show adversaries automating multi-stage intrusions using publicly available AI frameworks at unprecedented speed.
What's happening
- Attackers used Cairn, a publicly available framework from the Chinese open-source security community, in an aerospace sector intrusion observed in the last quarter of 2026.
- Cairn and similar tools including ARTEX, CyberStrikeAI and PentAGI enable AI agents to share information and autonomously decide which hacking techniques to deploy, executing intrusions at greater operational velocity than manual approaches.
- Chinese operatives used AI to translate, analyze and structure stolen information into intelligence reports; one attacker's system was configured to study software patches, develop attacks based on learned patterns and execute them fully automatically without human involvement.
- North Korean hackers used AI-generated fake recruiter profiles to target cryptocurrency organizations in North America and Europe, then leveraged Claude Code configuration changes in company repositories to trigger malware downloads.
Why it matters
- The defense industrial base faces a critical vulnerability: AI-powered penetration testing frameworks are publicly available, lowering the barrier to entry and enabling threat actors to scale campaigns targeting aerospace and other sectors globally.
- Fully automated attack chains eliminate the operational delay imposed by human decision-making, compressing the time between initial access and system compromise and straining incident response timelines.
- Organizations must now patch vulnerabilities before attackers using automated systems can exploit them, shifting the burden of cyber defense to continuous, rapid patching rather than detective controls.
Going deeper
- Google did not identify the specific aerospace organization compromised, disclose what information was taken or attribute the intrusion to a particular government, leaving the scope and consequence of the breach undisclosed.
- The report shows AI is now embedded across the full attack kill chain: reconnaissance, lateral movement, data analysis and social engineering, with foreign intelligence services and cybercriminals treating AI as infrastructure rather than a tactical tool.
The intrigue
- Google observed a China-linked group developing its own automated hacking system likely for future spying operations, suggesting state-level investment in proprietary AI attack frameworks beyond public tools.
The fine print
- The intrusion timing is described only as occurring in 'the last quarter' with no specific date disclosed, and Google did not attribute the activity to a particular threat actor or nation-state.
- The finding of a fully automated attack chain executing without human involvement is noted as a first observation by Google researchers but may not represent the first such attack conducted by adversaries.




