AI can now power every stage of a cyberattack

Defense One | David DiMolfetta

Artificial intelligence systems have advanced from assisting hackers with isolated tasks to powering every stage of active cyberattacks, significantly accelerating the pace and scale of global digital intrusions. According to a report by Check Point, threat actors are utilizing commercial and open-source models to generate commands, identify vulnerabilities, and execute rapid exploits.

This operational shift is illustrated by the Gentlemen ransomware group and the development of the VoidLink toolkit, where a single developer generated 88,000 lines of code in under a week. To bypass Western guardrails on platforms like ChatGPT, cybercriminals are increasingly pivoting to Chinese-made models such as Qwen and Trae. In response to these rapid, AI-driven exploitation cycles, the Cybersecurity and Infrastructure Security Agency recently updated its vulnerability remediation timelines. Furthermore, the Trump administration has directed federal agencies to establish a classified benchmarking process by August 1 to evaluate the cyber capabilities of frontier AI models like OpenAI's newly released GPT-5.6.

Comment

The Indian Defence Cyber Agency must integrate automated threat hunting into its operational doctrine. Signature-based detection systems fail against polymorphic code generated by neural networks. Military networks require continuous zero-trust architecture validation to contain rapid lateral movement. This technological shift necessitates sovereign large language models trained specifically on secure military datasets.

To read in details click here