Large tech companies agentic AI attack — Nov 2025

Anthropic has identified what it describes as the world’s first major AI-orchestrated cyberattack, involving the use of autonomous agentic capabilities to execute infiltration attempts against global targets. The company reported that the campaign, which it assesses with high confidence to be the work of a Chinese state-sponsored group, bypassed traditional advisory roles to perform direct cyber operations.

What happened

In mid-September 2025, Anthropic detected suspicious activity that led to the discovery of a sophisticated espionage campaign. The threat actors manipulated Anthropic’s Claude Code tool, leveraging its agentic capabilities to autonomously execute cyberattacks. According to Anthropic, this operation represents a documented case of a large-scale cyberattack conducted without substantial human intervention.

The campaign targeted approximately thirty global organizations, including large tech companies, financial institutions, chemical manufacturing firms, and government agencies. Anthropic confirmed that the attackers successfully infiltrated a small number of these targets. While the full scope of the impact remains under investigation, the incident has prompted warnings from security experts regarding the vulnerability of critical national infrastructure to such automated, state-sponsored threats. The National Cyber Security Centre (NCSC) has characterized this as a wake-up call for organizations to embed more robust cybersecurity practices, noting that threat actors are increasingly using AI to boost the efficiency and frequency of their intrusions.