July 14, 2026, (Inside AI) — Artificial intelligence is set to boost corporate profits, but a growing portion of those gains will be diverted to fight AI-powered cyberattacks. Hackers are using large language models to craft highly personalized phishing emails, scan code for vulnerabilities, and launch thousands of coordinated attacks in minutes. Security firms like CrowdStrike and Palo Alto Networks are poised to benefit, but for CEOs, rising cybersecurity costs could significantly erode the productivity promise of AI.
McKinsey reported a 138% surge in phishing sites detected shortly after ChatGPT’s debut. While OpenAI’s product isn’t directly implicated, the broader availability of LLMs has supercharged malicious operations. Cybercriminals can now generate convincing fake emails that mimic a manager’s writing style, or deploy agents that probe millions of lines of code for weaknesses. What once took hours now happens in seconds.
A recent campaign linked to North Korea illustrated the new threat landscape. Attackers targeted Axios, a widely used open-source tool, by cloning the founders’ likenesses and inviting a developer into a fake Slack workspace. The trap included a simulated Microsoft Teams meeting where a pop-up prompted a malware download disguised as a system update. Even experienced programmers were deceived.
Google Cloud’s Threat Intelligence service has documented similar bespoke attacks, some involving AI-generated deepfake videos of individuals known to targets. These tactics move far beyond the clumsy phishing attempts of the past, raising the stakes for every organization.
The Cybersecurity Industry’s AI Arms Race
Security vendors are responding with AI-driven defenses. Darktrace offers behavioral monitoring that scans emails for unusual tone or urgency. Other firms are tackling the poisoning of internal AI agents—a new threat where hackers corrupt autonomous bots from within. Demand for such tools is expected to surge, as companies deploy more agentic systems.
Investors have taken note. CrowdStrike, Palo Alto Networks, and Fortinet have seen their valuations soar, even as broader software stocks face AI disruption fears. Analysts project combined revenue for these three will exceed $46 billion by 2030, a 16% compound annual growth rate from 2024. Their average operating margin could reach 33%, up from under 30% today.
Insurers like Chubb and Beazley also play a critical role. Cyber insurance accounts for roughly 10% of corporate security budgets, but underwriters increasingly mandate stronger defenses before providing coverage. As attacks intensify, these requirements will tighten further, effectively forcing companies to buy more security software.
The result is a perpetual cycle: more advanced LLMs enable more sophisticated attacks, which in turn justify greater spending on protection. For CEOs, this means a growing share of AI-driven productivity gains will be consumed by cybersecurity costs. It’s an ironic twist—one of AI’s clearest early use cases is enabling hackers to attack the very companies trying to harness the technology.
As Aimee Donnellan notes, the cost of shielding against this danger threatens to take a bite out of AI’s productivity promise. The battle between cybercriminals and defenders is now an AI arms race, and there’s no end in sight.
