In-House Counsel Confront AI Contract Pressures
In-house legal teams are experiencing pressure as businesses push to deploy new AI capabilities and buyers demand unfamiliar commitments, according to a discussion with John Pavolotsky, a technology transactions attorney at Stoel Rives. Pavolotsky advises drafting contracts based on the current regulatory landscape and potential shifts in the next six to twelve months, highlighting that legal adaptation is essential to keep pace with company demands.
The Shifting Regulatory Terrain
AI regulation forms a patchwork, with California featuring dozens of AI-related bills and the EU AI Act categorizing systems into risk tiers that affect U.S. companies. Pavolotsky notes that states act as laboratories for governance, experimenting ahead of federal frameworks, and in-house lawyers must build contracting strategies to withstand this volatility. For businesses, staying aligned requires ongoing internal discussions about AI design, deployment, and use, as the regulatory environment remains unstable for years.
Defining High-Risk AI Use Cases
High-risk AI applications are outlined in the EU AI Act and Colorado AI Act, including sectors like education, housing, financial services, and government services that impact livelihoods. According to Dealbreaker, the challenge for in-house counsel lies in mapping organizational AI use cases and ensuring product managers define high-risk internally, with procurement workflows flagging relevant systems before contracts reach legal teams.
Evolving AI Contracting Practices
AI is fundamentally software, but contracting for agentic AI systems—those that act autonomously—requires addressing new risks like delegation boundaries and failure modes, differing from traditional SaaS negotiations. Pavolotsky provides an example of an AI travel concierge that could book flights and coordinate vendors without human intervention, potentially upending existing contract clauses. According to Dealbreaker, legal teams must experiment with AI tools to understand their mechanics, such as drafting clauses and testing outputs, to prepare for real risks.
