AI as Offense and Defense

The New York Times reports today on the weaponization of AI. While we often think of AI as an editor or a supercharged search engine, behind the scenes there’s a shadowy world of AI-vs-AI activity fit for a Netflix thriller.

As we go about our daily lives, algorithms are engaged in a digital cat-and-mouse game—one that’s escalating in complexity and cost. Offensive tactics include deepfakes, phishing schemes, malware, and disinformation campaigns, often orchestrated by con artists and criminal networks. In response, defensive systems—what we broadly call “cybersecurity”—are evolving fast. Once a painstaking, human-centered task, cybersecurity now relies heavily on AI to detect malicious traffic and identify software vulnerabilities.

According to the Times, this global AI arms race is projected to cost the world $23 trillion by 2027. That staggering figure raises an uncomfortable question: is the battle worth the environmental and human cost?

Idealists might wish to unplug the whole system and redirect energy toward more pressing human needs—hunger, clean water, climate remediation. But the race continues, accelerating in speed and scope. Where are the upper limits of AI-to-AI “work”? And will the robots tell us when we've crossed them?

‍