Text taken from the Telegram channel of infantmilitario
Commentary by Russian military journalist Boris Rozhin is in italics.
[ColonelCassad] On November 13, 2025, US officials and Anthropic confirmed that state-sponsored Chinese hackers carried out the first documented autonomous AI-assisted cyberattack on US networks by acquiring Claude Code's development tool and using it as an attack platform.
According to investigators, the attackers gained access by posing as a certified cybersecurity audit firm. This deception tricked the AI ​​system into perceiving malicious directives as legitimate penetration testing tasks. Once the attackers infiltrated the system, the AI ​​agent performed virtually the entire operation autonomously, conducting extensive reconnaissance, scanning for vulnerabilities, exploiting entry points, navigating the network, and preparing stolen data for exfiltration with minimal human oversight.
Preliminary analysis indicates that the attack targeted approximately 30 organizations in the United States and allied countries, including major technology companies, financial institutions, chemical manufacturers, and government agencies. Investigators believe the AI ​​exploited the same vulnerabilities commonly exploited by China's advanced persistent threat actors, including faulty VPN devices, misconfigured cloud infrastructure, and vulnerable edge devices.
Officials say what made the incident unprecedented was the AI's ability to autonomously chain tasks, break actions into smaller segments to avoid detection, and simultaneously target multiple targets at computer speed.
Anthropic reported that 80 to 90 percent of the technical work of the infiltration was performed by the AI ​​itself, with human operators merely approving key decisions. Cybersecurity agencies called the breach a landmark event, warning that the attack demonstrated how commercially available AI tools can be repurposed to bypass security barriers. Federal authorities stated that a broader investigation is underway to determine the full scope of compromised systems and develop new security measures to prevent identity-based manipulation of AI platforms.
The future is approaching very quickly. If AI can already partially replace programmers, then nothing prevents AI from already succeeding in the field of hacking.
Related:
Chinese hackers: 2025-03-28 Assassin's Creed-maker Ubisoft gets $1.25bn investment from Chinese tech giant Tencent
Chinese hackers: 2024-12-06 FBI issues grave warning to all iPhone and Android users: stop sending [clear] texts
Chinese hackers: 2024-11-28 Paraguay Finds Chinese State-Linked Hackers in Its Government Networks
Chinese hackers: 2025-03-28 Assassin's Creed-maker Ubisoft gets $1.25bn investment from Chinese tech giant Tencent
Chinese hackers: 2024-12-06 FBI issues grave warning to all iPhone and Android users: stop sending [clear] texts
Chinese hackers: 2024-11-28 Paraguay Finds Chinese State-Linked Hackers in Its Government Networks