Governed autonomy, enforced in hardware.

AI systems act only within verified authority. Hardware‑enforced. Fail-closed by design. Sovereign Australian.

GOVERNANCE LEDGER · DEMONSTRATION SEALING
The runtime enforces; this is the evidence it produces. Sealed, ordered, verifiable. Demonstration stream.
Hardware-anchored · Deterministic · Cryptographically accountable
3 Australian provisional patents filed Tested against a FIPS 140-2 Level 3 certified cloud-hosted HSM TRL-6 runtime / TRL-7 hardware modules Engineered in accordance with FIPS 204 100% Australian-owned

00 · Problem

“Human in the loop” is a policy document.
It is not a control. We make it one.

Every framework, standard and regulation assumes someone is accountable for what an AI system does. SCYTHEON makes that assumption enforceable: authority is verified at the hardware boundary, or the action does not execute.

01 · Sequence

Four stages. Every action.

Every governed action passes the same sequence before it executes, or it does not execute.

01

Policy check

The action is evaluated against policy before anything executes.

02

Authority verified

Human authority is proven at the hardware boundary, not assumed.

03

Decision sealed

The outcome is committed to a tamper-evident record.

04

Link verified

Continuity is confirmed. Any rewrite breaks the chain. Any gap fails verification.

ENFORCEMENT BOUNDARY ACTION + SIGNED AUTHORITY EXECUTE ACTION · UNSIGNED NO ACTION
ENFORCEMENT BOUNDARY · FUNCTION MODEL · DEMONSTRATION

02 · Platform

The SCYTHEON runtime

A patent-pending AI governance runtime anchored in hardware. Four functional layers, one guarantee.

Enforcement Boundary

Authority verified at the hardware boundary, not the application layer. Designed so that governance cannot be bypassed by the system it governs.

Evidence Ledger

A tamper-evident record of every governed decision. Cryptographic accountability built for audit, review and scrutiny.

Human Authority

Human-override safeguards engineered for high-consequence actions. Authority is a verified credential, not a checkbox.

Sovereign Operation

Air-gapped by design. No cloud, no callback, no foreign dependency for core governance.

> No signed authorisation = no action.

03 · Capabilities

Target environments

Deterministic governance, cryptographic accountability and hardware-anchored enforcement, designed for cyber-physical, autonomous and regulated AI systems that require safety, explainability and control.

Autonomous systems and robotics
Architected for deterministic validation of autonomous platforms. Hardware-enforced safety boundaries.
Command and control
Architected for governed command-and-control decisions with accountable evidence records.
DDIL environments
Designed for deterministic governance of platforms operating where connectivity is denied, degraded, intermittent or limited, so enforcement continues when the network is gone.

04 · Validation

Verification status

Prototype demonstrated in a relevant environment; hardware governance modules tested at software and synthesis level.

Tested FIPS 140-2 Level 3 HSM Governance modules tested against a FIPS 140-2 Level 3 certified cloud-hosted HSM. Integrated tests passing.
Signatures Ed25519 + ML-DSA-65 Classical and FIPS 204 post-quantum signature support.
Verification Formal checks passing Machine-checked verification of governance properties.
Maturity TRL-6 / TRL-7 Product runtime TRL-6; hardware modules TRL-7 (software/synthesis).
Detailed technical evidence, including validation results, test coverage and technical documentation, is available to qualified evaluators under NDA.
Deployment targets: Air-gapped · On-premises · Sovereign cloud · Tactical edge

05 · Sectors

Built for Australia's demanding sectors

AI governance for organisations where control, auditability and resilience are non-negotiable.

Defence

Command, control and autonomous systems in contested and denied environments. Architected for DDIL conditions.

Critical infrastructure

Energy, water, transport and telecommunications operators meeting national resilience and prudential obligations.

Government

Decision support designed to produce evidence fit for parliamentary and public scrutiny, keeping determinations audit-ready.

Financial services

Audit-ready evidence for algorithmic risk and operational resilience.

Robotics, maritime and aerospace

Designed for deterministic autonomy of robotic and autonomous platforms. Hardware-enforced safety boundaries.

06 · Contact

Request a briefing

A capability statement is available on request.

Public domain email addresses (such as Gmail, Hotmail, etc.) will not be accepted.
Your details are used solely to respond to your request.