Justin E. Pierce, co-chair of the intellectual property division of Venable LLP, addresses the unique liabilities and risks that are emerging with the growing use of agentic artificial intelligence.

There are dozens of legal cases pending in the U.S. right now concerning intellectual property ownership and the use of copyrighted material, caused by the arrival of agentic AI. The autonomous nature of AI agents — their ability to take actions with minimal or no human intervention — raises a number of questions that need to be resolved.

Creative content that’s generated by purely autonomous AI agents isn’t protectable by copyright, Pierce says. Only if a human steps in and makes some meaningful modification to the agent’s results does copyright law apply. That can be a concern for companies that are deploying agentic AI, yet view its output as proprietary and a source of competitive advantage.

Existing legal frameworks for address tort and product liability aren’t entirely adequate to settle the issues that are arising from the use of agentic AI, according to Pierce. “That’s the gap we’re trying to cover right now,” he says. Agentic AI is still in the nascent stage, and its output can be dynamic and unpredictable, potentially infringing on protected or personal information. What’s more, “it’s not 100% accurate at all times. Mistakes can happen.”

Nevertheless, Pierce says. Individuals who claim to have been harmed by the actions of agentic AI, especially where a human failed to constrain or supervise it, will likely seek to hold the company in question liable for damages.

Pierce says companies must focus on proper governance, including the segregation of sensitive data, to ensure that they’re deploying “responsible” AI, and involve stakeholders from multiple functions in exercising proper oversight.