AI is changing cybersecurity on both sides of the firewall. As attackers automate more of their work, defenders are beginning to use AI agents to continuously attack their own systems, find what gets through and test whether the fix actually works.

WHAT’S HAPPENING

Cybersecurity teams have traditionally relied on human red teams and breach-and-attack simulation tools to test whether their defenses can stop real-world threats.

That process is beginning to become more autonomous.

A new generation of agentic breach-and-attack simulation systems can use AI to select attack techniques, test security controls, identify weaknesses and help verify whether fixes actually close the gap.

Picus Security’s 2026 Blue Report analyzed more than 338 million simulated attacks conducted across customer environments during the first half of the year.

The report found average prevention effectiveness of 69%, while 58% of simulated attack activity was logged but only 14% generated alerts. (picussecurity.com)

WHY IT MATTERS

Cybersecurity has traditionally depended on people discovering a threat, recreating it, testing defenses and deciding what needs to change.

AI agents could compress that cycle.

Instead of waiting for a scheduled security test, companies may increasingly run systems that continuously ask:

Can this attack get through right now?

WHO BENEFITS

Security teams could identify weaknesses faster and spend less time manually recreating new attack techniques.

Companies could also test whether security tools are actually working instead of assuming that installed protections are enough.

WHO LOSES

Organizations with outdated security controls may discover that simply buying more cybersecurity software does not guarantee protection.

Security teams could also face new risks if autonomous testing systems are given excessive permissions or poorly controlled access inside production environments.