· ai-devtools

Sophos finds Aur0ra ransomware gang used AI agents for network access and credential theft

Artifacts recovered from the Cursor AI coding assistant in late August revealed that attackers were tasking AI agents with various operations, including being provided with VPN logins, proxy access…


Artifacts recovered from the Cursor AI coding assistant in late August revealed that attackers were tasking AI agents with various operations, including being provided with VPN logins, proxy access, specific targets, and defined objectives. This evidence comes from Sophos, the firm that monitors the ransomware collective known as GOLD OGRE. Sophos reports that the group responsible for the Aur0ra ransomware has been leveraging AI agents to navigate compromised networks, gather user credentials, and ultimately escalate privileges to attain domain administrator access. The process of this AI-assisted intrusion workflow was detailed in a publication that initially appeared on Help Net Security, titled "Ransomware gang talked AI agents into doing much of its dirty work."

Why it matters

This demonstrates a significant evolution in threat actor tooling, leveraging AI to automate complex attack stages. Developers must be aware that sophisticated malware is now integrating AI capabilities into its operational workflow.

Earlier coverage

Related coverage

Also worth reading

  1. Angular 20.3.35 and 21.2.26 released, fixing downgraded component template compilation
  2. next.js 16.4.4 released
  3. KloudStax CTO details first ten lines of command policy for AI coding agents
  4. Byoky, open-source wallet, stores OpenAI, Anthropic, Gemini AI API keys on device
  5. India floats AI plan under which ‘The model travels to the data, rather than the data travelling to the model’