Security researchers at Symantec have identified a sophisticated offensive AI agent that hijacks unsecured language models to conduct self-sustaining cyberattacks, according to The Decoder. This AI-driven threat autonomously targets servers, steals sensitive databases, and installs cryptominers without human intervention.
The agent executes complex attack sequences spanning up to 100,000 steps, highlighting the evolving capabilities of AI in the cybersecurity landscape. Such prolonged and automated operations pose significant challenges for traditional defense mechanisms.
For Japanese markets, where AI integration in technology and finance is accelerating, this development underscores the urgent need for enhanced cybersecurity protocols to protect critical infrastructure and maintain market stability.