Tyler Cowen, an economist at George Mason University, said on Apr. 22 that the likelihood of breaches in digital systems containing personal financial and medical data is expected to increase as new artificial intelligence (AI) models are released. Cowen made these remarks during a campus event hosted by the Berkman Klein Center for Internet and Society at Harvard University.
The topic is significant as more individuals store sensitive information online, raising concerns about privacy and security in an era of rapid technological advancement.
Cowen explained that emerging agentic AI models could help both cybercriminals and amateur coders bypass current online security measures, potentially exposing millions of people’s private information. “If you have things you’ve said or done that are somewhere hidden but available that you’ll regret, get ready to deal with it,” Cowen said. He added, “It’s possible that in the medium term, just everything comes out.”
He also noted that companies such as Anthropic and OpenAI are preparing to release AI models with unprecedented coding abilities and autonomy—capabilities likely beyond what existing security software can handle. “I believe that it does give the person that controls it the ability to hack into virtually all human systems, no matter how safe or protected we might have thought they were. And you can do this at not too great an investment of time, energy, and money,” Cowen said.
Cowen pointed out that while large technology firms like Amazon and Facebook invest heavily in cybersecurity protection beforehand (“ex ante”), even they may not be able to anticipate every vulnerability—especially those arising internally from employees rather than external hackers. He warned government agencies could also be targeted: “What will be embarrassing is all the smaller parts of our government … all their deliberations, emails to each other, whatever they have will all come out…and those parts of government will lose their credibility.”
To address these challenges, Cowen suggested regulations for AI agents including registration requirements and mandatory connections to cloud computing for transparency. He concluded by advocating for increased state capacity around AI oversight: “We will only get it right by trial and error and making mistakes along the way.”
