API
< Back to all clusters
[TECHNOLOGY] · Taiwan, United States · 8 sources

started · updated

AI-driven cyberattacks escalate against servers and infrastructure

Cybersecurity researchers and government agencies are reporting a significant escalation in the use of agentic AI to automate and enhance cyberattacks. Threat actors, including Chinese-speaking groups like UAT-10147, are utilizing AI agents to conduct reconnaissance, research vulnerabilities, and execute exploits against internet-facing Windows and Linux servers. These operations allow for much faster attack lifecycles, with some frameworks capable of autonomously rewriting code to bypass security detections.

Specific campaigns have been identified, such as the use of DeepSeek and the Hermes Agent framework to automate attack sequences. In the APAC region, near-autonomous AI frameworks have targeted government networks in Taiwan. Additionally, the SilkParasite operation has been observed using AI-driven polymorphic engines to change malware signatures in real-time, making detection by traditional antivirus software difficult.

In response to these evolving threats, US security agencies, including CISA, the FBI, and the DOE, have issued warnings regarding AI-generated exploits targeting critical infrastructure, specifically industrial control systems like Siemens S7 controllers. Experts emphasize that the risk stems from the combination of high AI autonomy and insufficient technical boundaries, necessitating a shift toward machine-speed defensive capabilities.

Entities

CISA · Cisco Talos · DeepSeek · FBI · SPECTRE · Taiwan · UAT-10147

Claims

What the coverage asserts, and how many sources carry each claim.

  • [○ 1 SOURCE] US security agencies warned that AI-generated exploit scripts are targeting critical infrastructure, specifically Siemens S7 controllers. www.ad-hoc-news.de
  • [● 2 SOURCES] A linked server held a target list of approximately 170,000 URLs. cybersecuritynews.com · nationalcybersecurity.com
  • [○ 1 SOURCE] A Chinese-speaking threat actor used DeepSeek and the Hermes Agent framework to automate reconnaissance and exploit acquisition. cybernoz.com
  • [● 2 SOURCES] The threat actor UAT-10147 uses AI-assisted tools to target internet-facing Windows and Linux web servers. cybersecuritynews.com · nationalcybersecurity.com
  • [○ 1 SOURCE] A near-autonomous AI attack framework has been deployed against government networks in Taiwan and the APAC region. flagthis.com
  • [○ 1 SOURCE] The group UAT-10147 used the SPECTRE backdoor for Windows and Linux systems. nationalcybersecurity.com
  • [○ 1 SOURCE] The SilkParasite espionage operation uses AI-driven polymorphic engines to rewrite malware code every few hours. que.com