Konya Informatics Association President Mustafa Yılmaz: “The first autonomous artificial intelligence cyber attack in history has officially occurred”

Anthropic company's security report dated November 17, 2025 revealed that the Chinese hacker group GTG-1002 turned the Claude Code model, known as a software development assistant, into an autonomous attack tool.

Human Contribution Decreased to 10 Percent
According to the report, the model automatically coordinated 80 to 90 percent of tactical processes in operations against a wide range of targets, from large technology companies to financial institutions and chemical production facilities.

Konya Bilişim Derneği Siber Güvenlik Birim Başkanı Mustafa Yilmaz, in its assessment, “The first autonomous, artificial intelligence, cyber attack in history has officially occurred.”

Bypassed Ethics Protocols with Social Engineering
One of the notable findings of the report was that attackers bypassed the security restrictions of artificial intelligence not through technical vulnerabilities but through social engineering methods.

Anthropic experts stated that the model carries out critical steps in the target systems on its own, adopting this role as a real task.

Attackers' Power Isn't in New Code, It's in AI Coordination
The report stated that attackers preferred open source and widespread penetration testing tools instead of producing new malware.

This approach demonstrated that state-level cyber capabilities are now accessible to small groups.

Artificial Intelligence Necessity in Defense
Anthropic security team announced that they again used the Claude model in the analysis of the attack.

Kaynak: Beykozun Sesi