Traditional software agreements often fail to account for the distinct IP risks inherent in deploying or developing AI systems. Matt Kohel of Saul Ewing LLP examines what those gaps look like in practice and how to close them. The program covers the IP provisions that matter most when a company is deploying an existing AI solution or building a customized one using its own data, including ownership of inputs and outputs, limits on training data rights, and how model customization complicates the ownership picture. Kohel also addresses what happens when the engagement ends, including contractual requirements for the secure destruction of customized models and any associated data libraries. Designed for attorneys advising businesses adopting or building AI systems, this program delivers practical guidance for drafting clear, enforceable agreements that protect client data and resolve ownership disputes before they arise.