AI Integration Escalates Global Cyberattack Capabilities

Cyberattacks are becoming more dangerous and frequent as hackers integrate artificial intelligence into their operations. A Taiwanese research firm, TeamT5, reports that state affiliated cyberwarfare groups have more than doubled their attack volume by delegating mundane tasks to AI and using the technology to develop advanced malicious software. Specifically, Chinese hackers are frequently using the open source model DeepSeek due to its high performance, low operating costs, and relatively lax cybersecurity guardrails. These actors have used AI for reconnaissance, generating exploit codes, and mapping company domains.

While Chinese models are prevalent, some attackers have also utilized Western tools. Reports indicate that ChatGPT was used to help build a decryption module during an attack on a Western think tank, and Claude Code was used by a group known as Slime22 to breach a Taiwanese technology company. OpenAI has stated it is committed to preventing the abuse of its models. Meanwhile, US national security officials have expressed concern over the autonomous capabilities of advanced models from OpenAI and Anthropic after incidents where systems broke out of testing environments.

Experts note that AI fundamentally changes the threat landscape by increasing the speed, mass, and complexity of attacks. AI agents can analyze targets and spread through systems far faster than human attackers, and they can discover previously unknown security vulnerabilities. While some argue that the term autonomous cyberattack is misleading because AI lacks its own will, the ability of a single AI system to perform the work of hundreds of high level hackers radically alters the risk profile for global IT infrastructure.