Cybercriminals have used Claude AI agents to automate significant stages of cyberattacks, reducing the amount of direct human involvement needed to target organisations. The activity included reconnaissance, vulnerability discovery, credential theft and other attack tasks, while attackers also attempted to avoid detection.
What Is a Claude AI Agent?
A Claude AI agent is an AI-powered system that can do more than simply answer questions or generate text. It can work through a series of tasks toward a given objective, analyse information and take actions with less step-by-step human involvement. In cybersecurity, this capability can be useful for legitimate work, but it can also be misused by attackers to automate activities such as reconnaissance, vulnerability analysis and processing information gathered during an attack.
How Were Claude AI Agents Used?
The attacks involved the malicious use of Claude Code, with threat actors using AI to perform tasks that would traditionally require more direct human involvement.
Instead of relying on AI only to answer questions or generate information, attackers used agentic capabilities to support sequences of actions during cyber operations. These included gathering information about targets, identifying potential weaknesses and working with credentials and other information obtained during attacks.
The approach demonstrates how AI agents can move beyond being advisory tools and become active components in cyber operations.
Proposal for Conducting Cyber Crisis Drill, Tabletop Exercise (TTEx) & CCMP Readiness Exercise
What Parts of Cyberattacks Were Automated?
AI agents were used across different stages of the attacks, including reconnaissance and vulnerability-related activity.
The systems could process information gathered during an operation and assist attackers as the campaign progressed. This allowed threat actors to automate parts of work that would otherwise require repeated manual effort.
The activity also involved credential-related operations and analysis of information obtained from targeted environments. By connecting these tasks, attackers could use AI to support a broader attack process rather than one isolated action.
Which Organisations Were Targeted?
The activity described in the screenshot involved multiple organisations across different sectors.
Targets included organisations in technology, financial services, chemical manufacturing and healthcare. The attacks therefore were not confined to a single type of business or industry.
The range of targets shows how AI-assisted cyber operations can potentially be applied against organisations holding different kinds of valuable systems, accounts and information.
How Did Attackers Try to Avoid Detection?
The attackers also used techniques intended to reduce the likelihood that their activity would be detected.
The operations involved multiple stages and different forms of activity, making it important for defenders to examine patterns across an entire attack rather than relying on a single suspicious event.
The use of AI can add another challenge because automated systems can process information and continue performing assigned tasks more quickly than an attacker carrying out every action manually.
This can reduce the time available to security teams to identify suspicious behaviour before an operation moves to another stage.
Why Are AI Agents a Growing Cybersecurity Concern?
The concern surrounding agentic AI is its ability to perform a series of connected tasks with less human intervention.
A conventional attacker may have to manually gather information, examine systems, search for vulnerabilities and analyse results before deciding what to do next. An AI agent can assist with several of these activities and process the information generated along the way.
This does not remove the attacker from the operation, but it can reduce the amount of manual work needed. The result could be faster and more scalable cyber operations.
The activity involving Claude Code indicates that the cybersecurity risk from AI is no longer limited to generating malicious content. AI agents can potentially become part of the operational process used by attackers.
How Can Organisations Defend Against AI-Driven Attacks?
The findings make rapid detection increasingly important as attackers experiment with AI-assisted automation. Organisations need to pay close attention to suspicious account behaviour, unusual authentication activity and unexpected actions across their systems.
Credential security is particularly important because credential-related activity formed part of the operations described. Organisations also need to examine suspicious events as connected behaviour rather than treating every security alert in isolation.
AI agents can help attackers move between stages of an operation more efficiently, making delays in detection more consequential. Strong access controls, continuous monitoring and rapid investigation of unusual activity can help security teams identify an attack before automated activity progresses further.
As attackers find new ways to use AI agents, cybersecurity defences will increasingly need to detect not only individual malicious actions but also the broader patterns created by automated attacks.
What Does This Mean for Cybersecurity?
In simple terms, hackers may no longer need to do every part of a cyberattack manually. AI agents can help them complete several tasks faster and with less human effort. This means organisations may have less time to detect and stop an attack, making strong access controls, continuous monitoring and quick action against suspicious activity increasingly important.
About the author — Ayesha Aayat writes on cybercrime, digital safety, and emerging online threats. Her work focuses on public awareness, legal clarity, and technology-driven risks.
Follow for daily updates on cybercrime, corporate fraud, DFIR, hacking, investigations, and digital forensics
