AI turbocharges China-backed cyberattacks, warns TeamT5
Chinese state-backed hackers have quietly weaponized AI, turning everyday large language models into force multipliers for their campaigns. According to Taiwanese firm TeamT5, the number of attacks traced to these groups has more than doubled since they began feeding exploit code and network reconnaissance tasks to models like DeepSeek, ChatGPT and Anthropic’s Claude Code.
The shift is visible in real-world incidents: hackers now automate the drafting of phishing emails, refine malware payloads and probe corporate networks faster than ever before. TeamT5’s analysis shows that once manual phases of an attack—such as vulnerability scanning and exploit generation—are now handled by AI, slashing the time from days to hours.
The new toolkit on the dark side
Open-weight AI models are no longer just research curiosities; they’re becoming off-the-shelf tools for adversaries. TeamT5 points to DeepSeek and other open models as key enablers, while Western models like ChatGPT and Claude Code are also being repurposed despite their safeguards. The rapid maturation of these capabilities is underscored by a recent UK study cited by the firm, which finds that open models are catching up to proprietary systems in cyber operations.
Who’s in the crosshairs?
Targets span critical infrastructure, tech exporters and government contractors across Asia and beyond. The automation allows attackers to cast a wider net, probing multiple sectors simultaneously with minimal human oversight. Security teams now face a dual challenge: detecting AI-generated threats and mitigating the accelerated pace of attacks that AI enables.
Why it matters
This isn’t just another cyber alarm—it’s a glimpse into how state actors operationalize AI at scale. For defenders, the doubling of attacks signals that signature-based tools and human-led incident response are no longer sufficient. Organizations must invest in AI-native detection, continuous monitoring and rapid patching cycles to keep pace. The window to adapt is closing; the question isn’t whether AI will reshape cyber warfare, but how fast defenders can catch up.
Source: The Decoder. AI-assisted editorial synthesis — TechnoExpress.