China Hackers Test AI-Optimized Attack Chains in Taiwan

Proofpoint and OpenAI observed DropPitch, a Chinese state-aligned actor, running persistent phishing and backdoor campaigns in 2025 targeting Taiwanese semiconductor investment firms, finance, academia and politically sensitive organizations; the group experimented with ChatGPT to generate multilingual phishing emails and to refine a custom backdoor ('HealthKick'), but produced low-quality artifacts and ultimately reverted to more traditional, person-to-person infection tactics.