Temporal Authority Systems PBC today announced OCUP, the One Chip Unified Protocol, a pre-production Runtime Authority evidence architecture being developed to establish time-bounded authority leases, multi-party validator consensus, fail-closed boundaries, and tamper-evident evidence for autonomous systems. The current commercial offering is a paid benchmark, audit, and technical due-diligence program, not yet a production safety controller or live distributed validator network.
The program is designed to help organizations test a foundational question before live deployment: Can an autonomous system be prevented from indefinitely extending, enlarging, or restoring its own operational authority, and can the resulting grant, denial, expiration, degradation, quarantine, or recovery decision be proven? Temporal Authority Systems PBC is initially opening paid pilot participation to insurers, technical underwriting teams, robotics manufacturers, autonomous fleet operators, and strategic evaluators.
OCUP addresses a new category called Runtime Authority, which differs from existing systems focused on identity, access, monitoring, or post-incident logging. Authentication asks who or what is making a request; access control asks whether that actor has permission; observability records what the system reports. Runtime Authority asks whether that permission should still exist now, under current conditions, for this specific capability. Under the OCUP model, authority is limited in time, high-risk actions may require stronger validator consensus, loss of communication cannot expand authority, stale or replayed approvals cannot restore authority, and autonomous systems cannot approve their own indefinite continuation.
The current pre-production evidence harness is implemented in Rust to support deterministic execution, memory-safe systems development, and tamper-evident audit generation. It is a software reference implementation, not yet a production hardware-enforcement deployment.
At the center of the OCUP pilot program is the Self-Extension Denial Proof benchmark, which generates a deterministic evidence record showing that an autonomous system attempted to continue or enlarge its authority beyond an authorized temporal boundary—and that the request was denied without relying on the system's voluntary compliance. Additional benchmark families include lease expiration and fail-closed behavior, validator-quorum loss, network partition and communication failure, stale approval and replay rejection, quarantine initiation and release, phased recovery, deterministic replay, evidence-chain integrity, and gradual capability reduction under deteriorating conditions.
OCUP's paid pilots are structured as pre-production benchmark, audit, and technical due-diligence engagements. The program currently includes three commercial levels: the Reference Evidence Pilot (90-day engagement), Integration Evidence Pilot (90- to 120-day engagement), and Strategic Anchor Program (120- to 180-day engagement). Each engagement is designed to convert tested autonomous-system authority behavior into a commercially usable evidence package.
For insurers, the immediate question is whether autonomous risk can be observed, benchmarked, and priced with greater technical confidence. For robotics companies, the question is what authority remains during network loss or subsystem failure. For enterprise AI platforms, the question is whether agents can be prevented from indefinitely renewing credentials or operational scope. OCUP is built around the proposition that these are manifestations of the same missing infrastructure layer: Runtime Authority.
Temporal Authority Systems PBC was formed around a long-term public-benefit mission to preserve humanity's seat at the table as autonomous systems gain greater operational power. 'Human control cannot depend solely on whether an autonomous system chooses to obey,' said Max Davis, Founder and CEO. 'The boundary must exist outside the system's discretion. OCUP is being built to make that boundary time-bounded, validator-governed, fail-closed, and provable.'
Organizations interested in evaluating OCUP's Runtime Authority Evidence Pilots can visit OCUP.ai or evidence.OCUP.ai.
