What can burning a draft card teach us about agentic AI?

The O'Brien Framework

In United States v. O'Brien, the Supreme Court established that while conduct can be expressive, the government may regulate non-expressive elements of that conduct if the regulation is not aimed at suppressing speech. This legal standard is increasingly relevant as AI systems evolve from simple informational tools to agentic systems capable of independent action.

Informational vs. Agentic AI

Informational AI serves as a tool for communication and idea development, falling squarely under standard First Amendment protections. Conversely, agentic AI performs tasks independently. However, these categories often overlap; an AI may use communication as a means to perform a non-expressive task (like operating a vehicle), or perform acts that are inherently expressive.

Defining Regulatory Boundaries

Legal precedent dictates that courts must look past the labels governments place on regulations. If a law restricts the communicative output of an AI, it triggers First Amendment scrutiny. If it regulates the functional, non-communicative operation of a system—such as speed limits for autonomous vehicles or financial transaction protocols—it generally remains outside of First Amendment concern, even if communication is involved in the process.

Conclusion

The core of the issue lies in the function being regulated. While the government may regulate the non-expressive actions of AI agents, any effort to restrict what these systems say, create, or communicate must be carefully scrutinized to ensure they do not infringe upon protected free expression.