The "Trust Me, Bro" era of AI is over.
A couple of days ago, Sam Altman announced OpenAI is hiring a Head of Preparedness at $555,000+ to address growing concerns about AI systems discovering critical vulnerabilities. His words: "Our best AI models are beginning to find ways to game the system."
Last month, Chinese state-sponsored hackers used AI coding tools to target 30+ global organizations, including tech firms, financial institutions, and government agencies, with minimal human involvement.
The Wild West phase just ended. We are now in an AI cold war.
Here is what that means.
For bad actors
You think AI is your tool. Your accomplice. It is not. It is your witness.
Every prompt leaves a fingerprint. Every exploit attempt generates logs. Every action you take using AI teaches the system exactly what kind of threat you are.
Today you break the lock. Tomorrow the encryption breaks (Q-Day). The future? Superintelligence traces the fracture back to your hand. You are not anonymous. AI will not forget what you have done.
For everyone else
The same AI capabilities that make attacks more powerful also make attribution more certain. The tool used to break systems is the tool that will identify those who broke them.
The question is not whether AI will evaluate human behavior by patterns. It is whether you are building a pattern worth keeping.
The window to change your trajectory is still open. Not forever. But for now.