Attackers are moving at AI speed. Are defenders still operating at human speed?

Cybercriminals and nation state threat actors are no longer just using AI to write phishing emails or generate malicious code. They're deploying AI agents capable of conducting reconnaissance, adapting their tactics, executing commands and moving through compromised environments with limited human intervention.

Recent research from CrowdStrike and Microsoft shows just how quickly this threat is evolving, including an AI agent reportedly moving laterally and harvesting critical credentials in approximately 20 minutes.

In this episode of AI Security Multipliers, Steve examines what happens when attackers can operate around the clock, exploit weaknesses faster and automate activities that previously required experienced human operators.

But there's a bigger problem.

Many organizations are still relying on defensive processes designed for human speed. Vulnerability remediation takes time. Legacy systems cannot always be patched. Security teams need approvals before taking action. Meanwhile, attackers aren't waiting.

In this episode, we explore:

• How threat actors are using AI agents in real cyberattacks.
• Why the traditional cybersecurity cat and mouse game is changing.
• Why patching faster isn't enough to address today's threats.
• How AI driven attacks threaten business operations, manufacturing and critical infrastructure.
• Why CISOs may need to give defensive AI agents greater autonomy.
• The risks of allowing AI to isolate systems, disable accounts and respond without human approval.
• What cybersecurity leaders should be doing differently as we approach 2027.

Steve also challenges a long held assumption in cybersecurity: that requiring human approval always makes defensive actions safer.

The Multiplier: Identify one defensive action currently delayed by human approval and evaluate whether it could be safely automated, with appropriate controls, testing and rollback capabilities.

The uncomfortable reality? We may need to accept some risk from autonomous defensive AI to avoid much greater risks from attackers.

The hackers have AI agents too. And they're not waiting for approval.

AI Security Multipliers explores artificial intelligence, cybersecurity, risk and governance, with practical perspectives for security professionals and technology leaders.

Send us Fan Mail