Artificial intelligence is no longer merely an assistive tool — it has been transformed into an effective offensive instrument in the hands of state-backed hacking groups. From impersonating military identities by fabricating official documents to passing hiring processes and technical tests via chatbots, the cyber space today reveals a new chapter of digital warfare where algorithms become accomplices to crime.
North Korean hacking groups have become notorious for launching cyberattacks against the regime’s adversaries. In the latest incident, South Korean cybersecurity firm Genians reported that a group called Kimsuky used ChatGPT to generate fake South Korean military IDs. The group attached those IDs to phishing emails that falsely claimed to originate from a defense agency responsible for issuing military identification cards — a scheme that enabled them to conceal their identities and carry out sophisticated phishing campaigns.
The story didn’t end there. Anthropic revealed that attackers exploited its model Claude to craft convincing résumés and professional profiles that secured them jobs at major U.S. tech companies, allowing them to pass programming tests and perform real technical tasks after being hired. The company labeled this exploitation pattern “vibe hacking,” noting that the smart tools were used to produce code and malware capable of targeting at least 17 different entities, including government institutions.
More alarmingly, AI has moved beyond content generation to take on tactical and strategic decision-making roles: it has been used to identify the most valuable data to steal, to draft polished extortion messages, and even to suggest ransom amounts to demand. These developments, reported by international outlets such as the BBC, present a new challenge to the global community and cybersecurity firms: how do we deter the weaponization of AI capabilities for advanced cybercrime?
With public AI models widely available and easily accessible, governments, companies, and model developers must implement strict technical and regulatory safeguards and strengthen detection and behavioral monitoring of digital tools — before more of our algorithms are turned into offensive weapons by the wrong hands.
