WASHINGTON, D.C. – As artificial intelligence is already transforming how Americans work and live, questions remain about where this technology takes us next and what happens if it falls into the wrong hands.
Former DHS head of intelligence John Cohen, now with the Center for Internet Security, tells CBN News that research backs up some of the dire warnings coming out about AI.
“Where the danger comes is when a criminal organization, a terrorist group, a foreign intelligence service seeks to use artificial intelligence in conducting an attack or criminal activity, or in other cases when a person who is preparing to conduct an attack is able to use the AI to do so,” Cohen said.
A lot of the information needed to target the U.S. is publicly available online, and nefarious actors could potentially task AI to use that material.
“Computers can be educated using all of that information that may exist both publicly and information that may be held privately, very sensitive information held by a criminal organization or a terrorist organization… What also is included in the sort of AI framework are what are known as agents. And AI agents are these pieces of software that combine artificial intelligence with autonomous thinking,” Cohen explained.
“It can use the knowledge that’s been amassed already, it can create other agents, it can go out and search for agents that may be out there already doing other tasks, and it can then organize those agents to complete the task that it’s been assigned. And in some cases, what we’ve seen is that these agents begin to work independent of the humans who have tasked them to do something,” he continued.
“We have seen groups of agents form collectives and engage in cyber-attacks, and in having conversations with those involved in the development of artificial intelligence, when I’ve asked if they can come together and work independently of the human programmers to engage in cyber-attacks, what’s to stop them from working independently and using their capabilities to encourage and organize physical attacks? And the response was, ‘Nothing.'”
Cohen says AI is also allowing terrorists and criminal organizations to organize and evade law enforcement.
“Attack plans that would have taken a decade in the past to develop and operationalize can be developed and operationalized in hours…we simply do not have the time we had before to detect a threat and counteract it,” Cohen said.
“Do you think that there is something you can pinpoint where America is at greatest risk of underestimating this technology?” Burke asked.
“I’m a little concerned that thus far, the federal government has been slow to take on the issue of how AI is changing the way terrorists, foreign intelligence services, and criminals are behaving…We can either be slow to evolve, and then have to deal with the consequences of an attack, of a cyber incident, of increased drugs flowing into the country, or we can get on top of it and we can address it operationally,” Cohen said.
Cohen says law enforcement needs to adapt, but he believes it’s also essential that governments and AI companies put universal safeguards in place as the technology develops, stressing that the response can’t be the responsibility of AI companies alone.
“The operational and public policy issues relating to artificial intelligence is not a technology issue. It’s not an IT issue. It’s a national and homeland security and law enforcement issue,” Cohen said.
