Something remarkable is happening in artificial intelligence. Some of the very people building the world’s most powerful AI systems are now saying we may need to slow down.
AI leaders are raising increasingly serious concerns about where this technology may be headed. They worry about AI systems becoming more autonomous, improving their own capabilities, and potentially operating in ways their creators didn’t anticipate. Whether you believe the most extreme predictions or think they’re overblown, there’s another question every organization should be asking:
What if we slow down AI—but the hackers don’t?
Jeff Lanza, a former FBI agent and sought-after cybersecurity keynote speaker, unpacks that question below—and what it means for the future of AI-powered cybercrime.
AI Has Already Changed the Rules of Cybercrime
For decades, cybercrime has essentially been a battle between attackers and defenders. Criminals find new ways to break into systems, steal information, and defraud people, while cybersecurity professionals develop better ways to stop them. AI has accelerated that arms race.
We’ve already seen criminals use artificial intelligence to:
- Create more convincing phishing messages
- Impersonate executives
- Clone voices
- Develop malicious software
- Automate parts of cyberattacks
But we’re moving beyond AI simply helping a hacker. AI is beginning to do more of the hacking itself. (For a deeper look at how these threats have evolved, see Jeff’s breakdown of the current cybersecurity landscape.)
From Chatbot to Orchestrator: How AI Is Automating Attacks
Recent threat research has documented cyber operations in which AI isn’t being used merely as a chatbot, but as an orchestrator—helping with reconnaissance, identifying vulnerabilities, writing code, and carrying out portions of an attack.
That’s a significant change. A hacker traditionally needed technical expertise, time, and patience. AI can provide all three. It can work faster than a human, operate around the clock, and give relatively unsophisticated criminals capabilities that once required highly skilled hackers.
Want your team to understand these threats before criminals exploit them?
Book Jeff Lanza for a keynote or webinar on AI-powered cybercrime, identity theft, and fraud prevention. Explore all of Jeff’s speaking topics to find the right fit for your event.
Why Slowing Down Responsible AI Won’t Slow Down Criminals
Now consider what happens if responsible AI developers decide their technology is becoming too powerful. They may slow development, conduct more safety testing, restrict certain capabilities, and build stronger guardrails around what their systems are allowed to do. Governments may impose additional regulations. Those may all be perfectly sensible precautions.
But there’s something I learned during more than two decades as an FBI agent: the good guys operate under rules. The bad guys don’t.
Law enforcement needs warrants. Financial institutions have regulations. Businesses have compliance requirements. Technology companies have safety standards and reputations to protect. Criminals have none of those constraints. In fact, criminals routinely look for ways to exploit the very rules and safeguards designed to protect the rest of us. I saw that dynamic throughout my FBI career, and we’re seeing a new version of it with artificial intelligence.
Cybercriminals and hostile governments don’t have safety committees or regulators looking over their shoulders. They’re certainly not going to stop using a technology because the rest of us have decided it’s becoming dangerous. And AI capabilities don’t simply disappear once they’re created. Models can be copied or modified, safeguards can potentially be circumvented, and powerful technology can spread around the world.
In other words, putting the brakes on responsible AI doesn’t necessarily put the brakes on criminal AI.
Could AI Safety Measures Accidentally Give Criminals the Advantage?
That creates a potentially troubling possibility. What if the companies developing some of our best AI technology—including technology that could help defend us from cyberattacks—intentionally slow their progress while criminals continue pushing forward? Could we unintentionally give the bad guys an advantage?
That’s not an argument against AI safety. If the people building the world’s most advanced artificial intelligence believe the technology is becoming powerful enough that we need to slow down, we should take that warning seriously. But cybersecurity needs to be part of that conversation.
Because there’s a fundamental difference between responsible organizations and criminals: we’re worried about what AI might become capable of doing. Hackers are thinking about what they can get it to do.
And they’re not likely to slow down just because we do.
Prepare Your Organization for AI-Powered Cybercrime
The threat landscape is shifting faster than most organizations can keep up with on their own. Employee awareness—not just technology—is one of the strongest defenses against AI-driven phishing, deepfake voice scams, and identity theft. See what past audiences have said about Jeff’s presentations in these testimonials, or check the speaker FAQs if you’re planning an event.
Want to help your employees, clients, or organization stay ahead of today’s AI-powered scams?
Contact Jeff Lanza today to schedule a keynote or webinar on cybercrime, identity theft, and fraud prevention.