“We Need to Be Able to Defend At Computer Speed”: Chairman Garbarino Joins Punchbowl’s ‘Fly Out Day’ to Talk AI and Cybersecurity

WASHINGTON, D.C. -- Last week, House Committee on Homeland Security Chairman Andrew R. Garbarino (R-NY) joined Punchbowl News' "Fly Out Day" to discuss how artificial intelligence (AI) is transforming the cybersecurity landscape and reshaping the resilience of America's critical infrastructure.

On the urgency of implementing U.S. frontier AI models in our cyber defenses, Chairman Garbarino said:

"We need access to these frontier models to be able to protect our systems… Anthropic came in and actually gave our committee a demonstration on Mythos. I have seen what it can do… They told the model to find a vulnerability in a bank and empty accounts. And it then went and did it. It also could then…find this vulnerability and fix it--and it did… Because the problem is Congress, as you said--we're very slow moving, and we're usually reactive and not proactive. So, I think they're showing us that to be like: This is where we are, guys. Like you can't wait for something like this to happen before you react, because that'll be too late… We can't continue to defend at human speed when they're going to be using computer speed to attack us. It's just not going to work out. We need to be able to defend at computer speed, and having access to models like Mythos to shore up our systems before there's an attack is very important. Anybody that would sit and go through and watch the demonstration of these models would understand that very clearly--that we need to act soon."

On the real-world impacts of AI on cybersecurity, Chairman Garbarino said:

"My number one concern, especially when it comes to AI and these models, is cybersecurity… We've had roundtables at the Committee… all these groups have come in, and they have not been open hearings. They've been amazing. I've had such great participation from members of both sides [who] come in. And, you know, you start talking about utilities, water utilities, power plants--that these systems were never supposed to be digital, but now they are. So you have possible vulnerabilities in those systems that might not be able to be fixed. And all of a sudden, you're going to find it out… It's one thing when information is lost or information is hacked. But it's another when the operation side is hacked. And now it's like, well, you know, we're going to turn off this gas pipeline or we're going to increase the amount of chlorine or whatever we're putting in the water. I mean, this is stuff that is scary."

On federal AI policy and the importance of AI safety guardrails, Chairman Garbarino said:

"I'm very happy that the White House did something with an EO. I'm happy with that. I mean, I'm happy that they included CISA in that as well. I appreciate Jay Obernolte and Lori Trahan coming out with something… I think federal pre-emption is something that should be part of it… I don't think we should slow down. I think we should be involved in testing. I think guardrails have to be in place… With these new AI models, you have the ability to not just learn how to do it, but to tell it to do it for you."

Background:

The House Committee on Homeland Security continues to engage with the private sector on AI through roundtables, briefings, and hearings on the future of cybersecurity and best practices for developing secure AI technology. Last month, Anthropic provided Committee Members with one of the first live demonstrations of Mythos given to Congress, allowing Members to see firsthand how advanced AI can find software vulnerabilities. The demonstration reinforced the urgency of ensuring that federal agencies, including our civilian cyber defenders, can responsibly use advanced U.S. models to find and patch vulnerabilities.

This month, the Subcommittee on Cybersecurity and Infrastructure Protection convened a hearing on AI and policy solutions to secure our networks, strengthen America's domestic technology base, and ensure the United States maintains its leadership in AI. Testimony was provided by Sandra Joyce, Vice President of Google Threat Intelligence; Dr. Chris Meserole, Executive Director for the Frontier Model Forum; Jack Cable, Chief Executive Officer and Co-Founder of Corridor Security Inc.; and Dr. Matthew Guariglia, Senior Policy Analyst at the Electronic Frontier Foundation.

In April, Chairman Garbarino and House Select Committee on China Chairman John Moolenaar (R-MI) launched a joint investigation into the growing adoption of artificial intelligence models developed by the People's Republic of China (PRC). The investigation is examining how the PRC is working to close the AI innovation gap with leading U.S. frontier labs through model distillation, intellectual property theft, and other illicit means, as well as the potential cybersecurity, data security, and critical infrastructure risks associated with the deployment of PRC-developed AI systems in American products and services. As an initial step in the probe, the Chairmen sent letters to Anysphere and Airbnb.

That month, the Committee also hosted a bipartisan, closed-door demonstration of "jailbroken" AI with presentations from the National Counterterrorism Innovation, Technology, and Education Center (NCITE). Their latest findings show a troubling trend: extremists are increasingly turning to "uncensored" or "unrestricted" AI models that remove built-in safety guardrails.

In December 2025, the Subcommittee convened a joint hearing to examine threats stemming from the use of emerging technologies like AI, quantum computing, and hyper-scale cloud infrastructure, as well as how to use this technology to defend our networks from malicious actors. Witnesses included Dr. Logan Graham, Department Head of the Frontier Red Team at Anthropic; Royal Hansen, Vice President of Privacy, Safety, and Security Engineering at Google LLC; Eddy Zervigon, Chief Executive Officer of Quantum Xchange; and Michael Coates, Founding Partner of Seven Hill Ventures.

###