Cybersecurity firms are increasingly deploying AI agents to counter more advanced attacks as hackers use AI for deepfakes, personalized phishing, and malware.

Brian Murphy, CEO of cybersecurity company ReliaQuest, said large enterprises face major challenges in detecting and responding to threats, with AI now helping to automate routine tasks previously handled by security teams.

ReliaQuest recently introduced autonomous, role-based AI agents that can take over functions such as detection and threat analysis.

Justin Dellaportas, chief information and security officer at communications technology company Syniverse, noted that AI agents now perform basic work like scanning logs, and can also take actions such as quarantining emails and restricting account access.

A May 2025 Gartner poll of 147 CIOs and IT leaders found 24% had deployed some AI agents, mostly for internal functions like IT, HR, and accounting.

Experts say AI agents are viewed as tools to support cybersecurity workers, not replace them, though concerns remain about their scalability and decision-making.

🔗 Source: CNBC

🧠 Food for thought

1️⃣ Cybersecurity follows predictable defensive-offensive technology cycles

The current AI arms race in cybersecurity reflects a pattern that has repeated throughout the field’s 50-year history.

When the first computer virus “Creeper” appeared on ARPANET in 1971, it was quickly followed by “Reaper,” the first antivirus program designed specifically to eliminate it1.

This cycle repeated with major threats like the 2017 EternalBlue exploit that enabled WannaCry ransomware, which prompted widespread security patches and new defensive protocols1.

Today’s AI-powered attacks follow the same pattern, with 70% of chief information security officers believing AI gives attackers an advantage, spurring rapid development of AI-powered defenses2.

The cybersecurity AI market is projected to reach $133.8 billion by 2030, demonstrating how defensive spending typically follows offensive innovation3.

2️⃣ Routine cybersecurity work faces systematic automation within four years

The automation of entry-level cybersecurity tasks is moving from experimental to mainstream deployment faster than many anticipated.