Agentic AI is arming cyber attackers with tools that plan and execute multi-step attacks on their own, and IBM's 2026 X-Force Threat Intelligence Index warns the escalation is real. The report found AI-driven attacks escalating sharply while basic security gaps continue to leave enterprises exposed, with AI helping attackers speed up reconnaissance, craft credible phishing and exploit publicly exposed applications at a pace defenders struggle to match.
What the IBM X-Force 2026 report found
- AI fuels attacks on basic gaps: the index links a spike in incidents to AI-enabled automation of credential theft and phishing against unpatched systems.
- Application exploitation is back in vogue: publicly exposed applications have become a leading entry point for attackers.
- Basic hygiene still decides outcomes: missing multi-factor authentication, weak credentials and delayed patching featured in most incidents X-Force handled.
- Attackers are faster and cheaper: agentic AI lets small criminal groups launch campaigns that once required skilled teams.
Why agentic AI changes the economics of cybercrime
Agentic AI systems can take a goal — such as compromising an organisation — and autonomously perform reconnaissance, select exploits and adapt when blocked. That compresses attack timelines from weeks to hours and lowers the skill bar. IBM's researchers note that AI does not replace human attackers; it amplifies them. Campaigns that once needed weeks of manual preparation can now be assembled in hours, and phishing messages generated by large language models are harder to spot because they rarely carry the spelling and grammar errors of older attacks. For defenders, the same tools can help, but only when security fundamentals are already in place.