AI Docket

What happens when artificial intelligence stops giving advice and starts taking action?

In this episode of The AI Docket, Velvet Johnson examines one of the emerging legal questions surrounding agentic AI: who bears responsibility when an AI agent takes an unauthorized, harmful, or unlawful action?

As organizations give AI agents access to enterprise systems, customer data, financial transactions, communications, and automated workflows, the legal risk changes. The concern is no longer limited to hallucinations, inaccurate outputs, or intellectual property issues. AI systems increasingly have the ability to act.

Velvet explores the accountability stack behind these systems, from model providers and platform developers to systems integrators and the enterprises deploying AI agents. She also examines the “three-party risk gap” that emerges when each participant assumes another party owns the risk.

The episode raises a question legal, compliance, technology, and business leaders need to address before deploying autonomous AI at scale:

When an AI agent acts on behalf of your organization, who is accountable for what it does?

The AI Docket examines the legal questions emerging as artificial intelligence moves from experiment to infrastructure.

Artificial intelligence is changing the nature of cyberattacks and exposing weaknesses in the legal frameworks designed to address them. As AI accelerates exploit development and lowers the barriers to sophisticated attacks, longstanding assumptions about intent, attribution, and accountability are beginning to break down.

In this episode of The AI Docket, Velvet Johnson explores how AI-assisted cyber operations are reshaping cybersecurity law. The discussion examines the Computer Fraud and Abuse Act, emerging questions about criminal intent and attribution, and the possibility that legal exposure could expand beyond hackers to the companies developing and deploying advanced AI systems.

The episode also considers what these developments mean for organizations building or adopting AI technologies and the governance measures leaders should be implementing today to prepare for tomorrow's legal and regulatory landscape.

Topics Covered

• AI-assisted cyber operations and the evolving threat landscape

• The Computer Fraud and Abuse Act, intent, and attribution in the age of generative AI

• Emerging theories of liability for AI developers and technology providers

• Governance, monitoring, and operational safeguards for enterprise AI

• Practical considerations for legal, privacy, cybersecurity, and AI leaders

In this episode of AI Docket, we explore a growing governance risk associated with AI systems: the potential loss of the model itself.

Algorithmic disgorgement reflects a shift in regulatory focus from data handling to system-level accountability. When organizations cannot demonstrate clear rights and provenance for training data, the resulting model may become difficult to defend.

This episode focuses on what leaders need to understand about data lineage, system design, and the conditions required to preserve AI assets under scrutiny.

In this episode, we explore the emerging legal risks of chatbot use in business settings, including how AI interactions are being treated in discovery and what recent court decisions signal about privilege and protection.

For leaders, the issue is not whether AI is being used. It is whether you understand the record your organization is creating when it is.

Artificial intelligence doesn't just process the data we provide — it can infer new information about us that we never explicitly shared. From predicting health conditions to estimating income levels or identifying behavioral traits, AI systems are increasingly generating personal insights based on patterns in seemingly ordinary data.

In this episode of The AI Docket, we explore the growing privacy risks associated with AI inference — one of the most overlooked issues in AI governance. Drawing on recent research analyzing more than 1,300 academic papers on AI privacy, we unpack why inference may pose a greater risk than the commonly discussed issue of AI memorizing personal data.

We also examine how regulators are beginning to respond, including how GDPR treats inferred personal data, how U.S. state privacy laws regulate profiling, and the emerging compliance challenges organizations face when AI systems generate predictions about individuals.

Finally, we discuss the practical implications for businesses deploying AI and the governance strategies companies should consider to mitigate these risks.

As AI becomes more powerful, privacy may no longer be defined solely by what we disclose — but by what machines can learn.

Building transformative AI? The gap between a working prototype and a legally defensible business tool is where most projects fail. In this inaugural episode, attorney Velvet Johnson and technical advisor Meghan break down the stark reality facing companies today: compliance is no longer an annual checkbox. It’s a daily operational necessity.

With over 20 state privacy laws, federal AI action plans, and cross-state enforcement actions, the regulatory landscape has become a living ecosystem that interacts with your technology every single day. Velvet shares real-world scenarios from her practice where brilliant AI models were halted because they were built on datasets that legally couldn’t be repurposed, and where unexplainable “black box” decisions became legal liabilities.

In this episode, you’ll learn:

∙ Why your AI inventory is non-negotiable and how it becomes your governance blueprint

∙ How to build auditable human oversight systems that satisfy laws like the Colorado AI Act

∙ The explainability crisis: why “the algorithm said no” is never a legally valid reason

∙ How to extend your compliance fortress through vendor due diligence and iron-clad contracts

∙ A three-pillar framework for building AI practices resilient to constant regulatory change

This isn’t about slowing innovation. It’s about building smarter foundations. The companies that systematically build trust through responsible, transparent, and well-governed practices are the ones that win long-term.

Perfect for: CEOs, CTOs, compliance officers, data scientists, and anyone building or deploying AI systems in regulated environments.