Aditya Vikram Kashyap Proposes Execution Sovereignty for AI Agents
Aditya Vikram Kashyap argues that sovereign AI must include execution sovereignty to ensure institutions maintain authority over what autonomous agents do and for whom they act.
AI governance leader Aditya Vikram Kashyap proposes that the framework for sovereign AI must expand beyond the traditional pillars of model, data, and compute sovereignty to include execution sovereignty. While existing sovereignty models focus on the physical location of infrastructure and data, Kashyap argues that execution sovereignty is required to maintain institutional authority over the actions of autonomous agents and the ability to withdraw that authority.
Kashyap warns that institutions risk being governed by foreign entities even while hosting data domestically if their AI agents rely on foreign policy engines or identity providers. He identifies early indicators of this concept in the European Union's AI Act and Data Act, which mandate the removal of switching obstacles and require human overrides for AI systems.
In the United States, the National Institute of Standards and Technology is addressing these concerns through its AI Agent Standards Initiative. The agency is currently researching identity infrastructure and agent authentication to determine the level of control institutions retain over systems acting in their name.
