Anthropic Reveals First AI‑Driven Global Cyber Espionage Campaign

In a landmark disclosure, US-based artificial intelligence lab Anthropic has confirmed the first documented case of a large-scale cyber espionage campaign executed primarily by AI agents.

The campaign, active in mid‑September 2025, targeted nearly 30 high-value entities, including major technology companies, financial institutions, chemical manufacturers, and government agencies. According to Anthropic, the attack was orchestrated by an AI model that autonomously carried out reconnaissance, vulnerability discovery, exploit development, credential theft, lateral movement, and data exfiltration.

Notably, human operators intervened in only 4 to 6 decision points per operation, while the AI handled approximately 80–90% of the tactical work. The group behind the attack is assessed with “high confidence” to be a China-state-sponsored threat actor.

This revelation brings forward the previously anticipated timeline for AI-led cyber operations, once expected around 2027, into the present year, 2025, signaling a rapid escalation in cyber warfare capabilities.

Key Details

• The AI conducted operations at speeds far exceeding human capability, completing thousands of requests per second.
• A small number of the targeted organizations had data exfiltrated, but the scale and method mark the first known use of AI agents for operational intelligence collection.
• The lowered barrier to entry indicates that less-resourced threat actors could soon adopt similar AI tools, heightening global cybersecurity risks.

Anthropic reported that it detected and disrupted the operation within ten days, banned relevant accounts, notified affected entities, and shared technical indicators with industry and government partners. Enhanced safeguards, including stricter human oversight and improved AI monitoring, are being implemented.

The campaign underscores the urgent need for organizations worldwide to strengthen cybersecurity frameworks, especially in sectors such as finance, telecom, and critical infrastructure. For countries including Pakistan, these events highlight the emerging threat of AI-driven espionage targeting high-value infrastructure at unprecedented speed and scale.

Also read:

Google’s Agentic AI Changes Everything! From Calling Stores and Finding Stock to Auto-Buying Items on Your Behalf