Re: AIVS Community Group: first-meeting call (proposed window May 22 to June 12)

Following my own ask, here's my intro.

I'm Erik Newton. I'm the founder of Sanctuary Framework, an open-source
sovereignty harness for AI agents (TypeScript MCP server; v1.2 just out on
npm). Sanctuary's core thesis is that human and agent sovereignty require
the same architectural primitives, and that these are foundational to human
flourishing. So, Sanctuary ships cryptographic identity, encrypted state,
approval gates, and portable receipts as a single coherent stack.

I'm also the author of Concordia (Python SDK on PyPI), an open negotiation
protocol for agents that includes a JWS-shaped receipt format. Full
disclosure: Concordia's receipt format overlaps with the AIVS problem
space. If the group considers it as a candidate work item, I'll defer to
member consensus on whether the chair should recuse from that decision; my
position is that chairs should not be selecting their own projects as
deliverables.

I'm also a co-founder of the California Institute for Machine Consciousness
with Joscha Bach.

What draws me to the verification layer specifically: the agent stack is
converging fast around capability, but the legibility and accountability
layers are lagging behind. Portable, self-verifiable cryptographic proof of
agent sessions is the keystone. Without it, every layer above (identity,
reputation, regulatory audit) reduces to vendor-attested claims. With it,
the substrate stays open and the operator stays sovereign.

I'm interested in standards work that furthers both human and agentic
sovereignty: legible records, portable data and identity, and auditable
behavior, all achievable without external infrastructure dependencies.

Best,
Erik


Erik Newton