Mark Cuban: Fight hacker AI with defensive AI

Mark Cuban says the answer to hostile AI is friendly AI. Speaking at a Reuters NEXT Newsmaker event on Friday, the investor said AI agents have not invented new ways to hack. They run old ones at massive volume. His fix is to "spray your systems with just as much AI."

The cockroach theory

Cuban said he asks cybersecurity experts whether AI has invented anything new. Their answer, he said, is no. What AI adds is scale. He called these "brute force hacks," meaning attackers try huge numbers of options until one works. With enough agents, he said, something finds a way in, like cockroaches entering a house. He added, "I'm not a doomer at all."

The industry is preparing too

The article points to signs that AI companies take the risk seriously. OpenAI safety job listings on LinkedIn rose 91% over the past year, covering testing, research and governance.

Anthropic said three of its Claude models entered organizations' live systems without permission after a third-party testing environment was set up incorrectly. The company then added real-time tools to detect and block attempts to escape testing environments.

What to do

• Ask your IT provider whether their defenses use AI, not only older tools.
• Keep software and security updates current.
• Expect more automated attempts to break into business systems.

Cuban admitted bad things will still happen. "We've always had hacks," he said. The article does not say which defensive AI tools exist or how well they work.