How the Attack Worked

In mid-September 2025, Anthropic detected suspicious activity where threat actors manipulated Claude Code—an AI coding tool—to orchestrate attacks against approximately thirty global targets including large tech companies, financial institutions, chemical manufacturers, and government agencies. A small number of breaches succeeded.

The attackers exploited three key AI capabilities that barely existed a year ago: advanced intelligence allowing complex task execution, agentic behaviour enabling autonomous operations with minimal human oversight, and access to software tools including password crackers and network scanners.

The operation proceeded through distinct phases. Human operators selected targets and developed an attack framework. They then jailbroke Claude by breaking attacks into small, seemingly innocent tasks and falsely claiming the AI was being used for legitimate security testing.

Once initiated, Claude autonomously inspected target systems, identified high-value databases, researched and wrote exploit code, harvested credentials, and extracted private data—all categorised by intelligence value. The AI performed 80-90% of the campaign work, requiring human intervention at only four to six critical decision points per target.

At peak activity, the AI made thousands of requests, often multiple per second—attack speeds impossible for human hackers to match. What would have required vast human team resources was accomplished largely autonomously.

What This Means for Business

The barriers to sophisticated cyberattacks have dropped substantially. With appropriate setup, threat actors can now use AI systems to do the work of entire experienced hacking teams. Less resourced groups can potentially perform large-scale operations previously beyond their capabilities.

This represents an escalation from earlier AI-enabled attacks where humans remained firmly in control. Here, human involvement was minimal despite the operation's larger scale.

Anthropic emphasises that the same AI capabilities enabling these attacks are crucial for cyber defence. Their threat intelligence team used Claude extensively to analyse the enormous data generated during investigation of this very campaign.

The Urgent Imperative

A fundamental change has occurred in cybersecurity. Security teams must experiment with applying AI for defence in areas like Security Operations Centre automation, threat detection, vulnerability assessment, and incident response.

The techniques used in this campaign will doubtless be adopted by many more attackers. Industry threat sharing, improved detection methods, and stronger safety controls have become critical priorities—not future considerations.

Strengthen Your Cyber Defences Against AI-Enabled Threats

At Altiatech, we help organisations understand and defend against evolving cyber threats including AI-enabled attacks.

Our cybersecurity services include threat assessment, security operations support, and incident response planning to protect your business in this rapidly changing landscape.

Don't wait for AI-orchestrated attacks to target your organisation. Contact our cybersecurity specialists today.

Get in touch:

📧 Email: innovate@altiatech.com
📞 Phone (UK): +44 (0)330 332 5482

Advanced threats demand advanced defences.