Forrester: Agentic AI Is Speeding Up Attackers, Not Changing What Stops Them

The fix for agentic-AI-accelerated attacks isn’t a new category of defense — it’s the oldest rule in security, Forrester analyst Paddy Harrington argues: an exposed, internet-reachable, unprotected system will be compromised, whether it’s IT or operational technology (OT). He points to recent attacks on U.S. water and wastewater facilities as evidence, where compromised devices “were visible… were reachable… became targets,” with no sophisticated technique required.

Two things are compounding that exposure, per Harrington. Automated discovery tools like Shodan continuously scan the internet for unprotected devices without needing to target a specific organization first. And agentic AI is now accelerating attacker reconnaissance and exploitation into faster, repeatable workflows — turning what used to require a skilled operator into something closer to an automated pipeline. His prescription is deliberately unglamorous: never expose OT systems directly to the internet, segment operational workflows from IT, and return to fundamentals rather than buying new tooling.

The economics behind that urgency show up in Jason Lemkin’s breakdown of Palo Alto Networks, in “5 Interesting Learnings from Palo Alto Networks”: CEO Nikesh Arora’s framing is that AI is raising cybersecurity’s terminal value rather than compressing it, since AI agents generate far more east-west network traffic requiring inline inspection. The company’s now running an $11.4 billion revenue rate, with its Cortex/XSIAM security-operations line alone growing 100% to $600 million-plus in ARR. BCG’s CEO-focused research, in “How CEOs Should Manage Escalating Cybersecurity Risks in the Age of AI”, puts a number on the exposure Harrington is describing: 35% of organizations report significant impact from AI-enabled attacks in the past twelve months, averaging three significant breaches and 25 sensitive-data incidents a year.

For sellers pitching into security or AI-adjacent budgets, the pattern across all three sources is consistent: the buying urgency is real and growing, but the credible pitch right now is disciplined fundamentals, not a new AI-defense category nobody’s proven yet.