Proving what an agent did: my Future of Privacy Forum panel
The open question is who holds the record once the agent acts.
Last month in Washington, I was on a panel at the Future of Privacy Forum's Annual Advisory Board Meeting, alongside senior counsel from IBM, Anthropic, Meta, and Medidata.
The framing was deliberately practitioner-focused: not how we think about AI governance, but what we are doing with these tools now. I used my time to walk through a working artifact rather than a framework, a Certificate of Action. It records what an agent was authorized to do, what it did, and the proof connecting the two.
The distinction I wanted to put before that group is straightforward. When an agent moves money, signs a document, or runs a workflow step, a policy tells you what should have happened, not what did. That difference is what surfaces in a dispute, an audit, or a regulator's inquiry.
The agenda arrived at the same place from another angle. One session examined how teams are bringing agentic tools into their risk assessments, and the honest answer in most organizations is that the policy is still being written. The open question is who holds the record once the agent acts.
Many of the people in that room are building the policy layer. Fewer are building the layer that proves it. That gap is the part I think deserves our collective attention.
