Agentic AI: Navigating the Uncharted Waters of Non-Human Identity and Liability

Agentic AI empowers automation and decision-making, but also expands the attack surface, raising novel legal and security challenges. This panel of cybersecurity attorneys explores the uncharted waters of non-human identity (NHI) and liability in the age of agentic AI. We will delve into the risks of compromised agents, identity attacks, tool misuse, and goal manipulation, examining potential liabilities under existing legal frameworks (Fair Credit Reporting Act, EU AIA, IP, tort, contract, product liability) in the absence of specific AI agent regulations. We will also analyze the OWASP Top 10 for LLM Applications and GenAI Data Security Best Practices, including Excessive Agency and Functionality, as well as gaps in existing threat-modeling frameworks for addressing the agentic AI attack surface, particularly when multiple AI agents interact within an ecosystem.