In this episode of AI Security and the Law, hosts Veronica Glick and Stephen Lilley speak with Amy Larsen, a technology policy and national security leader whose work has spanned across government, big law, and major tech companies, including Visa and Microsoft's Democracy Forward Initiative. They discuss how artificial intelligence is transforming the threat landscape for fraud, scams, cyberattacks, and disinformation – making them faster, cheaper, more personalized, and harder to detect. The conversation covers the blurring lines between attack categories, the rise of agentic AI and its implications for identity verification and liability, and the evolving patchwork of legal and regulatory responses in the U.S. and abroad. Larsen emphasizes that while existing legal frameworks provide a foundation, gaps remain – particularly around authorized push-payment fraud reimbursement, cross-jurisdictional coordination, and multi-agent transaction traceability. She closes with practical guidance for companies and individuals, urging tabletop exercises, crisis preparation, and consistent digital hygiene as essential defenses.

Show Notes:

(00:02) Introduction to AI Security and the Law (04:40) "AI Security" Blurs Fraud, Scams, Disinfo, Cyberattacks (06:51) AI Fraud Losses: $15.9B, up 28% YoY (10:15) Cheap Deepfakes, Real-Time Video, Payment-App Scams (13:13) AI Empowers Amateurs and Organized Crime Alike (15:00) Agentic AI Challenges Delegated Authority and Liability (22:22) Policy Gaps in Reimbursement, Enforcement, State Definitions (27:04) Tabletop Exercises, MFA, Phishing Awareness, LE Relationships

In this episode, Stephen Lilley and Veronica Glick sit down with Eric Wenger, Cisco's Senior Director for Technology Policy, to discuss how frontier AI models are reshaping vulnerability management. Eric shares insights from Cisco's participation in Project Glasswing and Project Daybreak, including how advanced models helped the company review 1.8 billion lines of code in eight weeks rather than an estimated eight years, while keeping false positives to roughly 3%. The conversation covers the tension between granting defenders' early access to cutting-edge models and getting innovative technology to market quickly, government initiatives like the AI vulnerability clearinghouse, and why fundamentals like multi-factor authentication and least-privilege access remain more important than ever. Eric closes with a grounded take: rather than fearing a science-fiction AI takeover, organizations should focus on accelerating patch cycles and shoring up the basics that have always mattered.

Show Notes: (00:02) Introduction to AI Security and the Law (01:53) From Cybercrime Prosecutor to AI Security Leader (05:25) The AI Inflection Point (09:20) Cisco's Role in the AI Security Ecosystem (10:50) Inside Project Glasswing and Project Daybreak (12:05) What Cisco Learned: Speed, Accuracy, and the Right Model for the Job (16:55) Rethinking Patch Cadence and Regulatory Expectations (19:45) Government Initiatives: The AI EO and Project Gold Eagle (23:00) Model Access, Scarcity, and Tools for Defenders (29:00) Closing Thoughts: Back to Basics

In this episode, Stephen Lilley and Veronica Glick sit down with Eric Wenger, Cisco's Senior Director for Technology Policy, to discuss how frontier AI models are reshaping vulnerability management. Eric shares insights from Cisco's participation in Project Glasswing and Project Daybreak, including how advanced models helped the company review 1.8 billion lines of code in eight weeks rather than an estimated eight years, while keeping false positives to roughly 3%. The conversation covers the tension between granting defenders' early access to cutting-edge models and getting innovative technology to market quickly, government initiatives like the AI vulnerability clearinghouse, and why fundamentals like multi-factor authentication and least-privilege access remain more important than ever. Eric closes with a grounded take: rather than fearing a science-fiction AI takeover, organizations should focus on accelerating patch cycles and shoring up the basics that have always mattered.

Show notes:

(00:02) Introduction to AI Security and the Law (01:53) From Cybercrime Prosecutor to AI Security Leader (05:25) The AI Inflection Point (09:20) Cisco's Role in the AI Security Ecosystem (10:50) Inside Project Glasswing and Project Daybreak (12:05) What Cisco Learned: Speed, Accuracy, and the Right Model for the Job (16:55) Rethinking Patch Cadence and Regulatory Expectations (19:45) Government Initiatives: The AI EO and Project Gold Eagle (23:00) Model Access, Scarcity, and Tools for Defenders (29:00) Closing Thoughts: Back to Basics

In this episode of AI Security and the Law, hosts Stephen Lilley and Veronica Glick sit down with Daniel Castro, President of the Information Technology and Innovation Foundation (ITIF), to unpack the latest executive order on artificial intelligence and cybersecurity. The conversation explores how AI is fundamentally changing the economics of cyberattacks, making it cheaper and easier for threat actors to exploit longstanding vulnerabilities at scale, and what the federal government's response means for critical infrastructure operators, state and local authorities, and private sector businesses alike. Castro offers insights on the executive order's three operative sections, including efforts to upgrade federal systems, the voluntary 30-day review process for frontier AI models, and the prioritization of criminal prosecution of AI-enabled cyber threats. The episode also touches on the broader policy landscape, including the bipartisan Great American AI Act, underscoring that this is a rapidly evolving area with significant implications for global businesses.

In this episode of AI Security and the Law, hosts Stephen Lilley and Veronica Glick sit down with Wei Chen, Chief Legal Officer of Infoblox, to explore how the rise of agentic AI is reshaping internet infrastructure, cybersecurity, and the practice of law. Wei explains how the Domain Name System—the internet's foundational routing layer—can play an important role in securing autonomous AI agents, and she discusses Infoblox's DNS-AID proposal to the Internet Engineering Task Force as a path toward open, decentralized governance of the agentic web. The conversation also covers practical advice for lawyers, including building technical fluency with AI tools and embracing disruption as an opportunity for growth.

In this kickoff episode of AI Security and the Law, hosts Stephen Lilley and Veronica Glick welcome cybersecurity leader Jeff Greene of Civira Partners—a veteran of DHS' Cybersecurity Infrastructure and Security Agency, the National Security Council in the White House, NIST, the Senate, and various private sector roles—for a wide-ranging conversation on AI and security. They discuss implications for businesses of AI models enabling vulnerability exploitation at unprecedented speed and scale; meeting the moment through rapid AI adoption while limiting security debt and associated legal risks; and why foundational cybersecurity practices and resilience remain critical even as companies manage distinctive threats to AI tools. While expecting the cat-and-mouse dynamic between attackers and defenders to continue, Jeff argues that organizations should prepare to cede more decision-making to automated defenses that operate at machine speed and adjust their cybersecurity programs to meet this new reality.