Chinese state-backed cyberattacks double with AI, says TeamT5

The report from TeamT5, a Taiwanese cybersecurity firm, illustrates a significant shift in state-sponsored cyber operations. The doubling of attacks is directly linked to the adoption of AI for routine tasks and malware development. The accessibility of models like DeepSeek appears to be a key factor, as its low guardrails make it an attractive tool for generating exploit code and other malicious content.

The use of various AI models—including ChatGPT and Claude Code—by different groups shows that this is not a single-tool phenomenon but a broader integration of AI into hacking workflows. The UK AI Safety Institute's study provides a comparative framework, suggesting that while open models are rapidly improving, they are not yet at the level of Western frontier systems for fully autonomous operations. This creates a temporary but distinct capability gap, offering potential defenders a window of time to develop countermeasures.