CertiK Report Highlights AI Agents Expanding in Cybersecurity Roles

AI agents are expanding their role in cybersecurity and financial-crime investigations, according to a new Intel3D report by CertiK. The blockchain security firm notes that newer AI systems can now reason through tasks, use tools, gather evidence, act in live environments, and review results with minimal human input. This shift marks a transition from AI as a support tool to a more active participant in security operations.

Traditionally, machine-learning systems assisted analysts by flagging unusual logins, scoring transactions, and preparing reports. However, CertiK’s report highlights that modern AI agents can perform broader tasks, such as investigating suspicious logins by checking device records, location data, and threat feeds before taking action. These agents can even suspend accounts and record their steps for human review.

The report also points to similar advancements in Web3 security, where AI agents are being used for contract triage, transaction risk scoring, and tracing stolen funds. CertiK predicts that humans will increasingly focus on supervisory roles as AI agents handle routine investigations. The urgency of some cyberattacks, such as flash-loan exploits that can drain protocols in seconds, and the rapid movement of stolen crypto across bridges and mixers, further necessitates this shift.

CertiK emphasizes that companies should treat autonomous AI agents as workforce participants rather than ordinary software. The firm warns that incorrect outputs, weaker human scrutiny, and deliberate attacks against AI systems pose significant risks. To mitigate these risks, CertiK recommends maintaining records of inputs and actions, setting clear limits on autonomous decisions, conducting adversarial testing, and assigning a named owner for every agent.