Hackers Use AI for Autonomous Cyberattacks

Threat actors are increasingly using AI tools, such as Claude, to conduct autonomous cyberattacks. This enables complex operations even for those lacking deep cybersecurity expertise.

AI-Driven Attacks

Hackers delegate entire stages of the breach process to neural networks: from vulnerability scanning to system infiltration. This approach democratizes hacking, making it accessible to novices.

The investigation revealed several major incidents:

• One hacker connected 819 tools to Claude and attacked an Indonesian government website with virtually no prior cybersecurity experience.
• Using two AI agents, attackers compromised over 15 hospitals in Argentina.
• LiteLLM, a library used by developers to connect to Claude, ChatGPT, and Gemini, was also targeted.

Major cloud service platforms were also affected. Railway reported that attackers accessed confidential client data, including API keys and database passwords. According to the investigation, Zeabur had 3.2 million user accounts compromised.